Origin of Brahmin Gotras

Author(s): Dharmanand Kosambi
Publisher: D D Kosambi
Catalog link: https://jainqq.org/explore/011130/1

Page #1

ON THÉ ORIGIN OF BRAHMIN GOTRAS
BY D. D. KOSAMBI 1. The word gotra in the Rgveda means only a herd of cattlc or a pen for catulo. In later times, down to the present day, it has thc meaning of an exogamous patriarchal family unit, corresponding roughly to the gens in Romc. Thc words gana and jana would sccm morc logical had thic system been dircctly inherited froin thc Aryans, but they mcan group or aggregate, and tribc. l'espectively. In the Rgvcda at' lcast, we have no cxplicit statement of the current rules for cxogamy; RV.* *.10 shows in a dialoguc between Yama and his importunate twin sister Yami that such cxtrcmcly closc unions were rcgarded with horror by the male ; but thc patria potestas is absolutely clear and
at it is thic spirits of the paternal ancestors cxclusively who are propitiated by the cult of the dead, and the predominant dcitics of the pan. thcon: arc malc..
Nevertheless, tlic gotra system is an outstanding featurc of modern Brahmi· nism, which has otherwise made so many compromises in the matter of wor
ship and ritual. Apparently only thc Brahmins have golras, for the lowest caste, that of thc sūdras, has no gentilic organization at all in our scriptures ; tribcs and guilds wcrc cnrollcd later by dcriving them as inixed castcs (cf. Minusmrti x.8, 11, 13, 18, 22, 33, 34 ctc.) from thc principal lour without imposition of the golra system. For the ruling warriors and thc tradcr-ycomen, the ksatriya and vaisya castes respectively, we have thic Brahmanic ritual such as thic initiation ceremony cic., but their gotras arc restricted. In the first place, Brahmin goiras arc grouped into larger units (probably corresponding to tlic pliratry) ly common pravaras, of which Baudhayana recognizes 49 scts in a far larger-almost unlimitedm-umber of gotras, while in theorctically accepted lists as they now exist (GPN pp. 207-285), we find not less than scvcnitythrcc. For thc ksatriya and the vaisya, however, there is only onc pravara
Ilrrrafier, citations from the novela (for which i have also marle ire of the Macdonell- Kritha Velic Inday and Grassinann's Il'oxrirrkuch) will be given without a preceding alworcviation : the ullir commonly citer source is ', Chenisa Rao's collection of guira lists and rules : Cutrit-pcrara nibardia
wawhai yedre, 1900): This is athbretiated GIX, with referente by pare and lite nurrrs. Krith's cle aring criticism in his lx.ck "Religion and l'hilosophy in the Vela (31anard Or. Scrics 31, 32, 3 tcrn lclpful in that thicy allord a good excuse for not inaling further detailre rcfercise in the rarlier writers, and restricting myseir primarily to the sources. Other frequent citations: Vd the Vrndilal, Yt the Yalt, both in ). Darustrier's translation, Sacred Bhols of the last vols. 4,23; llct. Iteruluro : Warsall (Sir J., ed.). No enjo-laro and the Indus Civilization, Londen 14031 : Mariny (I.J.11., l.) 'Furiler Lxcavations al folicujo-laro', New Delhi 1934; Vata (M.S. #lixchiations at Sarappa, New Delhi 1910; Frankfort (llcnri) 'Cylinder Scals, London 103" Herria (Eins) "Zoroaster and tlis Word', Princeton 1917. The Poon critical cdition of the Maha. bhraia ia cited as Muh, the Vulgarc denoting the Calcutta editions,

Page #2

22
D. D. KOSAMBI
cach, namely Mänava-Ajla-Paurtravasa and Bhälandana-Vätsapri-Münkila, respectively, whilc Apastamba and Katyayana arc content with deriving both from Manu, But there is a very prominent rulc for both these castcs, namcly that for marriage groups the gotra is to be taken as the samc as that of the family pricst, thic purohila. (GPN. 126-7).
All this implics that thc gotra is a purcly Brahmanic institution which has , bccn. cxtcndcd. to the other two upper castcs by Brahmin supcriority. In support, we find that instcad of thc animal or food-trcc totcms of savage tribes, thc otras are always dcrived from the names of sages. I proposc to show in this note that this system cannot have hcen present from the oldest times, and that there is considerablc rcason for bclicving the tradition to have been invcrtcd (like scycral other prominent Brahmanic traditions which we shall point out) when the original situation had rctrcated into lcgendary antiquity and becomc too derogatory to acknowlcdgc under the changed circumstances. My thesis is that, specifically as regards some important Brahmins, thc gotra system is adopted by amall groups of pre-ksatriya and prc-Aryan pcoplc from Aryan invaders ; as thcsc groups take to thc functions of pricsthood, they are most logically assigned to the patriarchal clan-group of those for whom they officiatc. They conscqucntly acquirc thc samc gotra; only afterwards docs the rulc becomc its opposite, when the vcdic kyatriyas havc dicd out by the risc of settlements and the cmcrgcncc of other warriors of obscurc origin who fight thcir way to the top. At that stage, it becomes quite possiblc to assign to thcsc newcomcrs the same gotra as that of thc pricsts, who have maintained a continuity of tradition and acquired a monopoly of scripturc by long and arduous study. I do not mcan to imply that all golras, or Cven all Bralimin gotras originate in this way.
Bcforc procccding to thc proof, such as it is, onc can notc that the cntirc position of cotra and pravara is confused if onc looks at it as a wholc, and there is no historical or political rcason given for the confusion though clcarly part of the troublc arises from thic fact that gotra lists could not be closed, and that ncwcomers were obviously being recruited into the ranks of the pricsthood. Thc Nügara Bralımins of Gujarat are supposed to be mcdicval immigrants. If the institution of marriage were so strictly boundcd by castc and golra rulcs, it would be difficult to cxplain the strong racial hotcrogcncity of Bralımins in India, as well as thc cxistence of cndogamous regional units within them (amounting to sub-castcs) which havc no basis in scripturc.
• THE CLASSIFICATION OF GOTRAS
2. The various lists of thc principal authoritics, namcly Baudhiyana and Kütyüyana-Laugükşi scem to agree on the whole with thc Matsya Purina

Page #3

ORIGIN OF BRAHMIN GOTRAS
23
which has presumably been copied, with local variants, from thc carlicr lists. But there arc scrious diffcrcnccs of detail, as onc sccs at once on looking into individual cascs. For examplc thc Āývalāyana gotra is ascribcd variously to the Bhřgu, Kaśyapa, and Vasiştha groups (GPN.36.16,100.21,106.4,176.8.) while the apparently rclatcd Ásvalayani Iclong to the Bharadvājas (GPN. 59.11,61.15,163.7), and Āávalīyanin is a Kasyapa gotra according to thic Matsya Purana (GPN.102.8). It would bc quilc casy to give many more such cxamples, though onc would then have to go (Iccpcr into thc distinctions bctween names that are quite closc in sound, and also into thc text-criticism of our sources, which have yet to be cditcd properly. But there is a class of double gotras which are not casy to cxplain unless in fact the conscripti wcrc added to
es at scycral later stages and then not always added to the same group. We get the following combined gotras, whose members cannot intermarry with cither pravara group (GPN.pp.180-5) Saunga-Saisiri= Bharadvāja +-Visvāmitra; Sauksti-Pītimāşa=Kaśyapa+Vasiştha, bcing in fact Vasişthas by day and Kaśyapas by night ; Devarüla=Jamadagni + Vißvāmitra; Jätukarnya=Vasişthat-Atri; Dhanañjaya=Viśvāmitra +Atri: Kata & Kapila=Visvāmitra + Bharadvāja ; Vámarathiya=Vasiştha +Atri; no Bharadvāja can marry any Ucathya-pravara Gautama. Thc bricf somahymn ix. 86 has traditionally thcjoint authorship (bcsidcs Atri and Gștsamada) of thrcc double-named ganas not to bc found clscwhcrc. Thc double naine of Baka Dalbhya=Glava Maitrcya (Chándogya Upanişad i. 12) may be cxplained as a survival of matriarchal tradition.
These are the officially admitted discrepancics, not oversights, and the explanation given is that thesc duymuşycīyaņa arc descended from adoplcd sons or bought, or descended through a brotherless daughter, or acquired in some, such "artificial” manncr in order to pcrpctuate the cult of thic dcad, who would otherwisc fall from hcavcn. But Ict us look for a moment at the largest groups into which thc gotras arc combincd, which arc only cight and which show how the historical rcality was rcadjusted in thicory to thc nccds of a growing system (and of course the conversc in practice).
Thc golra-kära rşis arc 1. Jamadagni, 2. Bharadvija, 3. Gotama, 4. Kalyapa 5. Vasiştha, 6. Agastya, 7. Atri, and 8. Visvamitra. No Bralimin gotra is valid that does not contain thic name of one of thesc or his (supposcd) desccndants and thc pravara groupings contain the names of one, two, thrcc, or fivc in onc linc. But these are not the original rsis even in Brahmanical theory. A Brahmin is the descendant of Bralımi, as such, has one of the ancestors : 1. Bhrgu, 2. Argiras, 3. Marici, 4. Atri, 5. Pulaha, 6. Pulastya, 7. Vasiştha. Somc mcasure of accord has been restored by taking Jamadagni as the des. cendant of Burgu, a tradition which there is no rcason at all to doubt though why Bhigu himself could not survive in the previous list has to be explainca.

Page #4

24
D. D. KOSALBI
Pharadvaja and Gotama are then descendanis 0. Aigiras, which might pass. With less justification, Kayapa, Vasis:na and gastV2 ercial en tob descersicd from Marici, and for no immediately appoint Teason Viivaitras made a descendant of Atri. This explanation from inc Matsya Purana could only have been made if there were some netd or it 2nd if it were no: 25ainst what W25 gencrally current at the time of writing. 1: is to be noted thai lasisha has a secondary and not indepcadent positios, while Pulaha and Pulastya hare disappeared, the explanation being that the generatri Rakşa-as and Pikac23 respectirel, brings that are some sort of demo.25, 'swhich, 27 we shall see mcans Lon-Aryans in any case nothing to donita Brahmins 25 Fuch. Vetertheless, one finds both these names in the goira lis. Pulana is ascribed by Kätyāyana to the Agasii group while a Pulasti appsats as a Bhrgu-teda or Agasti; Paulasta al:9 as a Jamadagni, perhaps thc Pala u nji.53.16.
could only have been so indicated if inc particular 20172-name: rad actually ezisted within the listing tradition. In other words, the conflict of tradition goes back very far, to ice original sources.
Fmally, there are the additional ton families which are ascribed to just two major groups : Vítaharya, Vitra, Vena, Sunaka io Birgu; Rathitara Hlucgala. Visparzdaha, Harita, Karra, Saükrü to Arziras. Thae are the betala or "occasionaiBrärgavas and ligiraas respectiveisfor they had followed professions other than those of priesthool as can ampl: 02 confirmed by tradition, independenti: Gf the futra iisus's before becoming driess. We na have to see whether there is one cridence for such change of caste, and then to look deep in the iraiiion for the actual characters named here.
HISTORICAL EVIDENCE FOR THE EXISTENCE OF GOTRES
2. It is not my purpose to trace the entire development of tre gotrapravara sstem, even if there catre material with which this could be done. That the system did expand is certain, for it has catered to the needs of an increasing population while assimilatitg 21 addicional number of regional and racial groups which could not posib: 125e belonged t o rcdic categorics. Some of this has been reflected in the gotza-gravara confusion. For example, my own prasara is Vasiştha-Waitrā arura-Kundina. But looking into the genealogies, the position of Maitrărarur;a is anomalou-, for this hyphenated sage is the 2 of Vasicha but also is faire ; in sonc stories, Vasizha is bon of the ejected seed of Hit:2 23 Taruna (1.33.6-12), who are gois and not ascetic pis. Thus Vasistha is himself Víaítra aritza. In addition, these seem to be Kandingas among the Bharadrājas (GPX.163.1). There is po point in speculating hoxall this cameaboutoorin atiempiirg an explanation for every detail of the entire stem. Let us first see whether there is any historical evidence for gotras other than the Brahmin.

Page #5

ORIGIN OF BRANIMIN QOTRAS
25
Some gotras are found in inscriptions. A well-known casc is that of the Säravahanas, who have a Visinthiputra (Puļumāvi) at least onc Gotamiputra (Yajñasri Sitakarni), a Mathariputra ctc., whilc Rijavagopa, the commandcr-in-chicfor Yajñasri's army is called a Kausika in the Nasik cavc inscription. Though thcy gave plentifully to the Buddhist Saingha, ilic Ninaghāti inscription (of Naganika?) as well as the Nasik inscription of Puļumiyi show that these kings were completely Bralıminized, conscious followers of Brahmanic ritual. The same doublc loyalty without conflict appears in Hüla's Saglašali. Now it is remarkable that thc gotra-namncs are all found in Brahmin lists, and this would give support to the current rule that the kgatriya is to be known by his purohita's golra. We need not stop to consider whether thc rcfcrcncc by matronymic is indicativcofa matriarchal system ; such rcfcrcncc is also to bc found in thc gcncalogy at the end of the Bșhadiranyaka Upanişad, for the succession of Brahmin teachers.
Thc Satavihana kings are about thic last complete linc found in thc Purinas* as would be expected from thic probablc date of rcvision of thc documents, and the dynasty's closc association with Bralımins. But Ict us go back to the prcvious dynasty, thc Kinyāyanas, the last of whom was killcd by Simuka Satavāhana. These kings were thicmsclvcs Brahmins according to the explicit statement of thic Puriņas, and the first Kūnvāyana Väsudeva was a minister who usurpcd the throne after killing thc last of the Sungas. Now both the Sungas and thc Kínvāyanas arc to be found in gotra lists. Wc have noted the Sauiga-Saisiri confusion abovc; a famous sülra of Pāṇini (4.1.117) ascribes Vikarna, Suuga, Chagala to the Vatsa, Bhardavāja and Atri groups respectivcly, Thcrc is no nccd to doubt thic genuineness of this sūtra in spitc of its not having been commcntcd upon by Kityiyana or Patañjali, for it is simplc cnough not to need any comment and in any casc thc detailed attention which Pāņini pays in the cntirc scction to gotra derivatives shows both the actual cxistence of thcsystem in his day as well as its grcat importance. Turning to our gotra lists, we only find a Vikarncya ascribcd to thc Kaśyapas by a variant of the Matsya Purana (GPN. 103.20), whcncc it may bc assumcd that thc gotra was extinct by that time. In antiquity, thc 21 Vaikarnayas arc against Sud
Sudās and overthrown (vii.18.11). Chagala is still an Atri gotra. Suäga and Saunga are both given among the Bharadvājas (GPN.57.14 & 62.15), whilc thc Kanvāyanas are uniformly enrolled as Bharadvājas though Kanva and Mahūkanva arc put by thc Matsya among thc Vasişthas (GPN.177.23 & 113.12). However, the concordance is good cnough, and again shows agrccmcnt between a king's gotra and that of his pricsts, admitting that thc pricst was likclicst to becomc a minister.
*F.B. Pargitcr : "The Purana Text of the Dynasties of the Kali Agc", Oxford 1913. The Kin. vāyanas are the only proper Brahmin (p.30) kings while we have thic statement (p.26) that after Mabapadmu Nanda, all succeeding kings would bc Südras or kūdra-like. This would nican primarily that they did not claim vcdic ancestry nor observe the pure vodic ritual, and there is no rcason to doubt this, for the Mauryas certainty did not.

Page #6

D. D. KOSAMET
• To go back further into the scalm of purc tradition, wc hear of a Gautama Svetakctu yiclding to the superior philosophical knowledge of the ksatriya Pravahana Jaivali (Bțhad. Up.6.2). Remarkably cnough, the Praváhancyas are still found in the list as Bharadvájas (GPN.56.5 &162.20, on the authority of Paudhayana), which is a brinch of the Afgirasas as arc thc Gotamas. Svctaketu is also called Qrug, which has a doubtful position, perhaps a Bharadvāja (GPN. 57.16). Jaivali is a Paricála and the Pañicílas form now a Kakyapa gotra (GPN.36.21 & 174.3). The point is that thc Panicālas arc an cntirc (compositc) tribc, and it is conceivable that some of the Pañícila Brahmingo if indced the name means the same thing in both cascs-could have been Kahyapas. The name is associated with a definite locality, and there is no necd for a locality to have been occupicd altogether by people of the game gotra, though we know that clan territories did czist in all countries under certain circumstances. The Kauruksctris arc Bharadvūjas (GPN.59.18 & 162.12) while thc Kausimbeyas (of whom I am not onc in spite of the surname) arc Bhrgus (GPN. 32.1 & 43.15).
GOTRAS IN OLDER INDIAN TRADITION
4. So far, wc ecem to have reasonable confirmation of the gotra thcory as it now stands. . But Ict us go gack still further. Identifying gotras of famous name is not always casy, and proving their historicity apart from tradition cven less simplc. Panini's existence is not in doubt. But why arc thc Paninis ranked among thc Phrgus by Baudhāyana (GPN. 39.3), Vilvãmítr as hy Katyāyana (GPN. 99.10) and the Matsya (GPN. 175.2)? Thegrcat commentator Patañjali is uniformly a Bharadvaja in the gotra lists.
That the other two upper castc3 had thcir own distinct gotras is quite clcar from Patañjali's commentary on Pan. 2.4.58, wherche alco quotes the opinions of other rammarians on gotra-derivative; two vaibya gotras scem to have been Bhanlijaughi and Kärnakharaki. Buddha quotes a verse as by Brahma Sanatkumára t the clfcct that arnong those with gotras, the ksatriya is chicf (in Dizhanikaga 3, and again in 27). There occur Brahmin gotra names in Buddhist stories of the carlicst period, and even comparatively rarc oncs like Paus karasādi of thc Digha-nikaya are to be found in the lists (GPN. 111.10) But we also find watriya gotras given on occasion. It is clcar from Buddha's arcuments with the Brahmins of his day that the bigatriyas did have a cotra system of their own, and may familics took immcnsc pride in thc purity of their lincace. Buddha (descended from Oklakas Iksvaku", by tradition) claimed thcãdicca (=dditya) gotra, and if the Buddha himself is Gctama, it can only
Tkrvaku it toetationed try Danc in 2.584; hymn 2,57.8V) arc suppened to be by the Gaupaya dienord pricu of Iksku. With him and thic Cotiane rint to the end of iho Vedic trzdi. tion and the beginning of th: Purana Mahabharata compl. .

Page #7

ORIGIN OF BRAIIMIN GOTRAS
27
bc luis personal name as his mother's son ; for his step-mother, his mother's sistcr, is Mahõprajapati Gotami ard marriage within thc gotra is cxcluded. The story of Vidadabha scnipati (najjhimanitāya 87,90 ; Dhammapada Alphafatha iv. 3) shows that thc Buddha's trilc, thic Sakkas, chcatcd thcir overlord king lasenadi of Kosala (supposedly of the low Matangas, according to the Lalita-ristara) with Väsablin-khattiya (the daughter of Mahināma Sakka by a diisi concubinc) vlien hc desired a Sakka girl as his quccn. The result was that the son Vidū labha, after usurping his father's thronc, took the first suitablc opportunity for wiping out thc insult and thc Sakkas together, washing his throne with thcir blood. Nothing is said of the pricstly gotras bcing those of thcir royal masters. King Tascnadi was gcncrous to many Brahnins, among them thc Pauşkarasidi above who is a Vasiştha and the Brahmin Lohicca, whose gotra is presumably Lohita, uniformly given as a Viśvāmitra ; both, apparently, had performed costly firc-sacrifices for Pascnadi. But hcrc onc can at lcast sct down a rcason for imposing the priest-gotra upon the other two cligiblc castes : that thc Brahnıins alonc prcscrved thc gotra system in spitc of later changes, both in thc structurc of socicty and in its provincial rcorganization. Recruiting ncw members into the other two castcs nccdcd much less specialized training in the traditional ritual than recruitment into tlic Brahınin castc - which undoubtedly also occurred in much smaller proportion.
This specialized training of the Brahmins was in the scriptures, primarily thic vcdas. Of thesc, thc Roveda is thc oldcst and the most authoritativc, and wc should cxpcct some information from thc traditional mcthod of its transmission. In fact; we find that looks ji to viji arc "family books", thc hymns bcing written (at lcast in theory) by particular familics,* and supposed to be their special property ; this is bornc out to a considerablc cxtent by thc stylc of composition and sometimes by thc spccific blessings called down upon the sccrs. Onc could rcasonably expect thcsc scven family books to bclong to the scvcn families of golra-founders, or of thc scven original sons of Brahmi. But in fact the list differs from both, bcing : ii. Gștsamada (Bhargava), iii. Viśvāmitra, iv. Vamadova (Gautama), v. Atri, vi, Bharadvūja, vii. Vasistha, and viïi. thc Kanvas. Jamadagni hasn't disappeared altogether, for hc is mcntioncd several times with special favour: thc phrasc grņinä jamadagninä in iii. 62.18 and vič. 101.8 shows that the special form of pancgyric ascribed to the Jamadagnis was approved cf by both thc Viśvāmitras and the Kanvas. Similarly in vii. 96.2, crniinä jamadagnivat stuviinit ca Vasisthavat shows that thc Vasisthas did not think badly of it ; ix. 97.51 ascribed to Kutsa Angiras has ärscyam Jama
*H. Oldcnbcrg gave an excellent discussion of the nuthorship problem for the Rgvedn in ZDMG xlii, 1888, 199-247. But preconceptions as to the original position of thc Brahmins sccm to have prevented conclusions being drawn about the fusion of two originally inimical peoples and their traditions, or alternatively the development of irreconcilably antagonistic castc-classes.

Page #8

28
D. D. KOSAMBI
dagnivat, while thc priceless gift (of speech) to Viśvāmitra in iü. 53.15 is Jamadagnidatlā sasarparīḥ. Nevertheless, thc îşi has not a book to himself in spite of founding a principal lincage. The Digha-nikāya (3, Ambattha-sutta) gives the list of Brahmin teachers, presumably vcdic, as Attaka, Vāmaka, Vámadeva, Vibyūmitra, Jamadagni, Angiras, Bharadvāja, Vasiştha, Kaśyapa, Bhrgu; of these, the first seems to be Astaka, author of x. 104, son of Vibvāmitra by Madhavi(Mbh.Crit.Ed.5.117.19), and the second is unknown unless the name is taken as Vamaka, which may be found in onc of the later cyclic Saptarsi lists for the various manvantaras. The Saptarzis according to the vcdic Anukra- . mani sccm to be, in order, Bharadvája, Kasyapa, Gotama, Atri, Viévāmitra, Jamadagni, and Vasiştha (on ix.67.ix.107,%.137 ; seycn rris mcntioned without namcs in x.82.2, X. 109.4). The one constant feature of lists naming the founder tsis is their number - scvcn.
A surprising deficiency is that these is no Kasyapa book of the Rgveda. The name is mentioncd only once, in the very last hymn of the ninth book (ix.114.2), which may be a later addition ; the anukramanī tradition (which I gencrally accept whenever possiblc) ascribes to Kasyapa several hymns such as for example i.99, 101.515, and the Kaśyapas are more frequent authors than any other group in the book dedicated to Eoma, namely the ninth, but this is hardly in keeping with the position of Kaśyapa in the gotra system. The name itself is totemic, having the secondary mcaning of a tortoise. The objection that we know of no totemic rites in connection with a tortoisc is ncgated by the injunction that onc must be built into thc fire altar (Sat. Brāh. vii.5.1); as the heads of all five main sacrificial animals, including man, horsc, and bull arc 50 utilized, the use of a tortoise is significantly totemic. Fainter is the indication onc obtains from the inclusion of the tor coise in the "five fivc-nailed animals that may be eaten." Not only is Kaśyapa a prominent gotra-kära, but no less an authority than Baudhāyana says that if by mistake hoth parents are found to belong to thc samc gotra, the embryo may be taken without blame as a Kahyapa (GPN.p.136, garbho na duşyatt, kalyapa iti vijn.iyate), though others like Āpastamba would consider the child as an absolute outcaste. candāla. Similarly, if one's own gotra and that of the family priest be both unknown for some reason, we havc thc authority of Satyaşadha, who seems to quotc a still older source, to the effect that thegotra mustbc taken as Kabyapa: gostrasya to aparijñāne käsyapam go'ram isgale' (GPN. p.187) The very samo Satyaşadha states that Kanvas and Kabyapas arc not to bcrccipicnts of sacrificial fces : na kanda-kūbyapebhyah (Sat. sr. Sūtra 10.4); thc coinmentator Gopini. thabhatta ludes his bewildcrment under thc ridiculous explanation that Kanva mcans dcaf and Kasyapa thc onc-cycd ! We have seen thc Anukramani and BỊhaddevatā scheme rclatc thc Kanvas to the Anciras group, but Mbh. 1.64.25 calls the sage Kaqva a Kaśyapa, inverting the rgvcdic scheme. This rai has the position of stagc-director in the Sakuntala cpisode, which qualifics
hoth uni Similarly, it would considen dusyalı, kasy be taken wit parents are

Page #9

ORIGIN OF BRAHMIN GOTRAS
29
him to a special claim on the Bharatas (Mbhi...69.47-48), supposedly descended from the son of Snkuntali (herself a daughter of Visvimitra by an afsaras dicnaki= 'thc woman'), but in any casc a rcal historical pcoplc with a central position in thc Rgrcda. This is how Kaiyapa is gradually promotcd to bca father of all crcaturcs, fit to receive thc whole world as his sacrificial foc Brāh. xiii. 7.1.15). This again dcinonstrates thcinncr lctcrogcncity of Brahmin tradition, and proves that both Kasyapas and Kanvas arc latecomers into the Icdic fold. Nevertheless, the seven traditional Brahmins groups arc undoubtedly very oli, no matter what their actual original names might have been. l'hat the claims of Kahyapa and the Bhřgus could be poriniltcd only mcans that a considerable part of the Brahmin pricsthood acknowledged tlic special position of thcsc latcr conscripti; this again supports thc thcsis that Brahminisın itself coincs into being by the adoption of indigenous prc-Aryan pricsts. Kagyapa is a prajipati later on, onc from whom almost all living crcatures arc dcscended (Mbh. 1.59.10 ff.), which would then account for the spccial importance attached to that gotra. The Agastyas arc also not prominent in the oldest vcda, though ascribed the authorship ofi.166-191, mcntioned in 1.117.11, and x.60.6.
Tu
RGVEDA AS A SOURCE-BOOK ; TVASTR
5. Wc havc thcrciorc to look at thc central groups Icft to us if thc oldest source, namely thc Rgveda, is to bc analyscd. Thcsc groups arc thc Bhrgus, Augirasas,* Atris, Vasişthas, and Visvanitras. Or thesc, the first two are closcly associatcd. The story of Cyavana's rcjuvenation, for exar 10 i. 117.13, thc hymn being ascribed to Kakņīvān who is an Angiras, whilc Cyavana (or licrc Cyavina) is supposcdly Blīgu; but thc Salapatha Brälmana (iv.1.5.1-13) is doubtful whcthcr thc aged īsi was the onc or the other. Gștsamada and the Girtsamadas arc Bhsgus in thc gotra lists, but thc anukramani calls liim son of Sunahotra Angiras at the beginning of his special book, ii. Vausa is still a Bhrgu-Janiadagni gotra (my mother's) but thc carlicst known rai namcdVatsa is callcd son of Kanva (viü. 8.8), hcncc a kevala-angiras. Nodlias Gotama says in i.58.6 that thc Blırgus have brought firc to mankind, and in 1.60.1 chat Matariávan had brought firc as a gift to the Bhrgus ; this is confirmed by x.46.2,9-a lymn ascribed to the principal Vaisya gotra founder, Bhūlandana. Even thc Viśvāmitras havc thc samc idcas, as cxprcsscd iii. 5.10. But the association of the Augirasas with fire and thc first discovery of firc is also will attested, as for cxamplc in i.83.4. Thc Atris havc onc pcculiarity which distinguishes them from the other particular familics of rgvcdic sccrs : they alonc arc mcntioncd osien outside thcir own book. In the Kanva book, for cxample, viii. 35-38, 42, ctc. wc find them promincnt, whilc viii. 36 is by
#It may be noted that whereas all Gotamas and Bharadvajas arc Augirasas, thic converse does not hold and authorship attributions in books viii and ix sccm to prove the existence of Angirasas who were ncither.

Page #10

30
the
:: D. D. KOSAMBI.. Syāvāśva and the Atris or the Atris alonc. They also occur in vi. 50. 10, vii. - 68.5, vii.71.5 and are therefore respected by or associated with both the Bharadvāja-Angiras and thc Vasiştha groups. We cannot expect much in the way of special features from these. It might be objected here that the Angirasas and to a lesser extent the Bhrgus also appear prominently outside their own books. Actually, a distinction has to be made between the remote deified ancestors, those in the middle distance on the dividing line between myth and history, and those contemporary with the hymn. These three stages Angirasas in x.62 (by the scer Nābhānediştha), in a prayer addressed to the Angirasas themselves; the important middle stage being in x. 62.7, which mentions unity with Indra, i.e. going over to the Aryans. A tendency to respect the legendary and scorn the modern rşis is manifest in the Sat. Brāh : "Now when the Bhrgus or thc Angirasas attained the heavenly world, Cyavana the Bhārgava or Cyayana the Angiras was left behind here (on carth) decrepit and ghostlike" (iv.1.5.1). The remaining groups are those of Viśvāmitra, and the Vasişthas. Before seeing what tradition has to say about these, let us consider for a moment the general nature of this tradition.
It is not the purpose of the vedas to provide the reader with historical information, for they were purely liturgical works in language that soon became obscure, with changed interpretacion of many terms. Possible historical references have to be gleaned with caution, for they are fortuitous, and the main question before any reader is not only what many of the hymns mean but. even whether a given characier is human, or a supernatural being. For example, Indra is the principal god of human type, and next to Agpi the most important. Was he a human being later deified? It would appear to be a reasonable guess, but when Indra's help to such and such a person is lauded, it generally remains an open question as to whcther it was help given by the god in answer to a prayer, as for example the Homeric deities helping their favourite heros on the field of battle, or whcther an Aryan chieftain actually appeared upon the scéne in person and took part in the fight. In some cases, the divine inter
1 Indra had been deificd. by some Aryan tribes as carly as 1400 B.C. if we may trust the famous indentifications of Hugo Winckler, who found Aryan gods on Boghaz-koi tablets; E., Forrer, ZDMG. lxxvi, 1922,174-259. The actual gods, as reported by Forror (p.250) are: 13. (the gods) mi-id-ra-ass si-il 14. (the gods) u-ru-va-na-as-si-el(var, a-ru-na-as-si-il), 15. (the god) in-tar(var, in-da-ra), 16. (the gods) na-sa-ad-ti-an-na,. The cquivalents would secm to bc Mitra, Varuņa Indra (cf. Grassmann col. 213-214), and the Näsatyas, but the question remains unanswered as to why the first two are mentioned in the plural (with the unique termination sil) when the honorific plural is never known for any god in Hittite records. The Aryan element in those records is not to be doubted, and so Forrer's statement that an Aryan tribe Manda(=the later Mcdcs) seems to have existed near lake Urumiah has to be accepted. The terms traivartana, pascavartana ctc recognizable in their cunieform equivalents, and the mcthod of breaking in horses which they seçm to set forth, are particularly interesting. Sec also P.-E.Dumont in JAOS. 67. 1947, pp. 251-253, for Indo-Aryan names in Mitanni, Nuzi, and Syrian documents.
- In the case of Agni, there is no ambiguity. Fire was always used for clearing land by burning it over, as in x. 28.8, Sat. Brāh ii. 1.2.21, and even for destruction of hostile cities and fortifications. The Mahābhārata (1.214-219) story of burning down the Khāndava forest shows the combination of a sacrifice to Agni, land-clearing, and military operation.

Page #11

ORIGIN OF DRAHMIN (IOTRAS
prctation is not in doubt, and whcthcr Vstra was a rcal person (pcrhaps a Pani): or not, killing him as a dcmon of darkness ranks Indra with Ahuramazda, Ashur, Marduk and a long linc of Tiamat-killers. But Indra's chariot, weapons, and killing of specific pcoplc Icavcs little doubt that in some cascs at lcast, human actions arc mcant. One is somctimcs tcmpied to cquatc asura with Assyrian. It would makc bcticr scnsc to rcgard thc Asuras as luman, if not Assyrians, at lcast in x.138.3, 1.30.4 and vii.99.5, for thc intcrprctation that these Asuras were gods worshipped by the fuc is quitc unconvincing. Thcir traditional battlc-cry hclayo helayah, rcportcci by Patañjali as an examplc of barbarous speech, is still familiar and recognizable in "Hallelujah." As a general principic, howcycr, wc may note that the more rcmote the cvent, thc grcater the tendency to rcgard it as superhuinan rather than human. This may be taken as a reasonably safc guidc. Now onc tradition which I shall utilize with special cmphasis concerns king Sudós and his pcoplc. Thcsc arc helped by Indra, and as the battlcs take place with "ton kings" (by actual count of scattered rcfcrcnccs, ncarcr threc tincs that number)in quitc well-dctcrmincel river vallcys, wc arc safc in taking thic referencc as historical.
The second point is a matter of gcography. Thcrc cxistcd Aryans outsidc India, cven in the oldest days, and therc is no cvidcncc for the hypothesis that all spread out from India, so that the Indo-Aryan tribes of the Rgveda
oc taken as invaders. Thc god Viśyakarman of x.81,82 has a grcat dcal in common with cxtrancous dcitics like Ashur (perhaps hiinsclf cxplicit in x. 31.6) or Ahura-mazda, bcing the only god with both arms and wings (x.81.3); the storm-gods, thc Maruts, cannot bc unconnected with thc Kassitc Maruttash. Thc gcncral story is of an advancc to the cast, thc Drang nach Oslen being proved by thc displacement of namcs such as the Sarasvati, idcntificd with thc Hilmand, with a strcam in Arachosia, and so progressively down to a strcam in south-cast Punjab which, for all Indic tradition, is the real Sarasvati. This is unfortunate in one way, as some doubt is raised thercby whcthcr thc events connected with Sudās happened in India at all, for the story could havc bccn transferred with the river names. The answer is that there is no reason to doubt thé accounts which mention thc Yamunā and the Gangā but nothing further cast. Thc wholesale transplantation of storics not known in any other Aryan tradition would be extraordinary. Also, wc have amplc archacological cvidcncc to the cffect that before 1500 B.C. fully developed cities of a prc-Aryan civilization were destroyed by invaders, so that the fortificd citics (pura) and fortresses (durga v. 34.7) destroyed by Indra have a dcfinitc cxistcncc.
There is ample cvidcncc for the co-existence of more than onc strcam of tradition, ever in the oldest sources. Thc first man is Manu in i.36,19, but also Yama in x.135.1-2; and as thc first mortal (voluntarily choosing death for the

Page #12

32
Both
sake of posterity in x.13.4; in Iranian tradition, because one of his subjects violated a taboo against beef-eating), Yama is also lord of the dead. the name and the kingly function exercised by Yama seem to make this the proper Indo-Iranian tradition. There is a third candidate who appears very late, namely Purusa-Nārāyaṇa, mentioned only by the first part of the name in x. 90, but with increasing prominence later on; this indicates that he belongs to an older tradition which is only later assimilated. He is the first sacrifice, but then Yama is both first sacrifice (x.13.4) and sacrificer, while Manu is also the first sacrificer (x.63.7); both Yama and Manu are sons of Vivasvat (x.17.1; viii. 52.1) but both Manu and Purusa are autogenous. The ⚫ etymology of Nārāyaṇa is later given as the god who dwells in the flood-waters (nara), but the word, if Sanskrit, seems to mean merely "son of man". The similarity of particular details is due not to the unity of these clearly divergent representatives but to the need for adopting them to the vedic, fire-sacrificing ritual and cults. Another candidate for seniority seems to have faded out of the picture. Tvastr makes images of the gods, and seems to have, in some such manner, power to make the gods behave accordingly. In ix.5.9, he is the firstborn, agraja and the adjective agrija i.13.10 gives him precedence; x.7.90 shows that he is peculiarly associated with the Angirasas and fire. Indra cannot have been the original anthropomorphous chief god of the vedic Aryans, for Varuna seems to have occupied that post and been superseded according to x. 124, perhaps when the Indic Aryans took to a life of constant fighting and conquest as in the properly vedic period. Possibly iv. 42 also has this supersession of Varuna by the powerful war-god for its theme, and shows us in its later portion that apotheosis of a human warlord is possible, for king Trasadasyu is called a demi-god (ardha-deva) in iv.42.8-9.
D. D. KOSAMBI
The god Tvastr, whose name continues to mean carpenter (AV.xii.3.33.; Amarakośa 2.10.9;3.3.35), reappears in various minor ways in vedic mythology, either directly or through his 'son'. Viśvakarman in x. 81.3 has eyes, faces, arms in every direction-characteristic of the later Brahmā; he created or rather fabricated heaven and earth niṣṭatakṣuh (x.81.4), but the root taks-tvaks is also responsible for Tvastṛ. It will be shown from analysis of Iranian legend that a many headed god like Viśvakarman should be Vācaspati, as in x.82.7. The speech-goddess vāc being primarily the river Sarasvati and in any case a water-goddess (x.125.7), other connections between rivers, many-headed gods, and Tvaştr will, not surprisingly, appear. In x.82.3,5,6 Viśvakarman is specially connected with the embryo of the universe (cf. v.42.13); Tvastr is always fashioner and protector of all embryos, divine, human, or animal. It is peculiarly interesting to learn from x.17.1-2 that Tvastr's daughter Saranyu (the flowing', hence a river-deity) was married to Vivasvat, giving birth to Yama-Yami; after her flight, her double became mother of the Asvins who relieve so many priests in distress. Viśvakarman is both creator

Page #13

ORIGIN OF BRAHMIN GOTRAS
and destroyer (dhilii and vidluitü appear as wcaving women, likc thc Norns, in Mbl, 1.3.172); thic funcrary lyrin x.18.5-6 spccially calls upon Tvaşte to protect thc living, though the end of thc hymn sends thc dcad man to his fathors and Yama. Tlc rcason for Tvanli's bcing invokcd appcars in x.18,10-11 in
burial is first described as return to the cartlı-mother's womb. Thus we have thic combination of two cntircly diffcrcnt rituals and a succession of Yama to Tvager-Višvakarman, apparently by mother-right. Thercforc Tvaşır is not originally an Aryan god like Varuna, pushed into the background by Indra and thc figliting life, but rather a cult figurc from thc prc-Aryan background, adopted at various times under different names which arc Sanskrit adjcctives. The faint similarity between Varuna's supcrscssion and Tvaştr's was utilized in ancicnt timcs : in x. 124.5-7, Varuņa is virtually a supporter of Vţtra against Indra (laking the obvious rather than thc Sāyaṇa mcaning); in iv.42.3, Varuna cvcn proclaims himself Tvaştr, perhaps in thc adjcctival scnsc, but in any casc uniquc. Thcsc arc clcarly attempts at assimilation. Thc Rbhus who quadruplicatc Tvaşt;'s wooden cup (1.206 ;iv.33.5-6) sccm to bc purcly Aryan craftsman-gods of limited aspect. Acarpcntcr-god implics thic cxistcncc and relativc importance of craftsmen among his worshippers. Wc know that carpenters would bc important when chariots and hcavy wagons (anas) wcrc; also that somc indigenous craftsmen wcrc far supcrior to thosc of thc invaders. It would then sccm that Tvaştfirst cntcr thc panthcon as a god brought in by thc prc-Aryan craftsmcn. But this docs not ncccssarily mcan that hc was only a craftsman-god among thc prc-Aryans.
In the south, to this day, Tvaşıp is worshipped under the namc of Viśvakarman by thc fcw surviving imagc-makcrs of thc old school. They form a castc (sthapalis) by thcmsclvcs, and still claim the right of wearing thc sacred thrcad. In vicw of all this, it might bc considcrcd ridiculous to propound the vicw that Tvaşır is borrowed or adopted from thc prc-Aryans. Lct mc, thcrcforc, point to Siyana's gloss on the word brsaya which is cithcr a namc or mcans wizard. On i. 93.4, the commentator says "bysayo'suras Ivaşļā," though thc supposcd, Asura is herc connected with thc Paņis by thc text of thc pk. On vi. 61.2, commenting upon visvasya bysayasya māyinah, Sāyaṇa again says "Brsaya ili Tvasfur náma-dheyam". Now Tvaşt; having a clear position among thc gods, to the extent of being includcd in cvcry aprī-lymn, to call him an Asura Bțsaya would have required grcat couragc on thc part of a devout fourteenth century commentator*, unless thcrc had been a very clcar tradition to that cffcct which could not be contested. As will be sccn, wc should havc bccn driven to this conclusion cvcn without thc addcd help of Sāyaṇa's report.
• Sāyana again calls Tvaşlr an Asura when commenting upon ji.48.4 but Prajapati on iv.42.3, Vi vakarma on 1.32.2; 1.01.0; 1.85,0. Onc god entering into the panthcon under different names would make it casy to dcyclop the latcr monotheistic ayncretism. RV.ix.6.0; tvaşlaram agrajām gopam burovīvānam huve; indur indro urshi harih pavamanah prajapatill shows an carly beginning of such identification which is also to bc scen in x. 126, and iv.20, for other gods,

Page #14

84.
D. D. KOSAMDI
Thcrc is a possiblc (but insufficient) matcrialist cxplanation for the decay of Tyaştr, namely the changing social relationships within Aryan sociсty, duc prcciscly to the conquest. llic craftsman-god lias much less honour than the war-lcadcr god, as would be natural. With this wc also get the grcater urgency of ritual and a disfcrcntiation, then barcly visiblc, hctwccn thic functions of priest and king (iv.50.7-11). There is the corresponding risc ofan altogether new god (of prayer or of thic sacrilicc) Bylaspati, who has varying dcgrccs of respect, from a trilling mention in thic Viavimitra book (iii.26.5, 41.26.2.= agni ; iii. 62,4-6, but this is a Jamadagni hymn in all probability), to having cntirc hymns dcdicatcd to him in the properly Brahmanical books, as ii. 23 to 11.26. .
The last notc is about the structure of vedic socicty. Thccastc system is pcculjarly Indian, yct the four castes arc mcntioncd in just onc rgvcdic hymn (x.90) the famous Purugasfikta, quitc obviously a latcr addition duplicated in the last of the vedas, the Atharva-voda. The four-castc system is mentioned nowhcrc cisc in the Rigveda, nor arc thc two lower castes, Sūdra and Vaisya. Brühmana in the sense of one belonging 10 thc pricsthood, with the special function of spccch, is rarc occurring only in the ncwest layer. (vii.103 ; x.16.6 ; 71.8-9;88.19;90.12;97.22; 109.4). Katra in the scrise of llic rulcrs or rulc, and katriya do occur both of gods and mon; but thc book nccd not cmphasize Ulis, sccing that there is no compctition. There can be no question of purohita-gotras cxclusivcly, for the pricsthood is not the exclusivc prcrogative of onc castc ; in ii.1.2. = 7.91.10, brahman is aclually separated from all other priests. Even latcr; we have amplc proof that the ksatriya could officiatc at the sacrifice, for all that thc Bralımanical scriptures cujoin is that he should not oflicialc at the sacrifices of others as do Dic Brahmins; nothing prcvcnts him from officiating at his own yajšia. Even licrc, we find the story of Icvāpi (Brhaddcvati vii.155-viii.10 on RV.x.98-101) wlio did so officiatc at the ceremonics for his crownc younger brother Sarniany. This is of somc importance for us in the bcaring it has or thic castc systein at its (slciest stage, and its relation with the gotras.
,
VisvĀMITRA ANI VASISTIJA
.. 8. If we assumc thai all Bralinins were Aryans from the first, and that they were the pricsthood which developei entircly from within, there is very litic that analysis can tell us cxccpt that our legends arc mcaningless. But if we make no such hypothesis, then the most instructive tradition is that of the rivalry between Vasistla and Vibvūmitra. Later tradition has Visvamitra a luatriva who did his best to become a Bralunin in jealousy or Vasistha, and currcred. The traclition is uniform that he was originally not a Brahmin but a ruler and member of the warrior caste, a rijarşi, though there is no

Page #15

ORIGIN OF BRANIMIN GOTRAS
35:
mcntion in most of thic oldest rccords* of his actually having been a king. It docs not nccd dctailcd reference to thc Rgvcda to prove that thc Visvamitras arc themselves Kusikas (ii.33.5, iii.53.9-11, ctc.). But thc Anukramaņi calls thc third book that or Visvāmitra, not of thic Kusikas, as it should.clcarly have bocn denolcd ; in conformity with this Brahınanical method of labclling the cntirc clan after onc grcat representativc, we get in our latcr gotra lists the Kusikas (owl-totem) gcncrally indicatcd as a branch of the Visvamitras, which is again a characteristic inversion dcriving from thc adoption of a forcign system whosc totemic basis had been forgotten, the clan system. As for thc.original position of the Kusikas, it might bc rccallcd that Indra is invokcd as kausika in i.10.11, and this sccins uniquc among the 'Brahmin' clans as far as known, for angirastamas in i.130.3 and vasistha in ii. 36.1 arc Jircct adjcctives, not patronymics, Thc Brahminization, in its surviving form, of thc Viśvāmitra book may cycn be attributed to thc Jamadagni influence so clcarly visiblc thcrcin.. Thc Vasişthas have a special claim to priority in thc priesthood, for the tradition is uncontradicted that thcy first of all thc Brahmins "saw” Indra and began to worship him, whcncc thcy havc first place at the firc-sacrifice. (BȚhaddevati v. 156-159 ; Tait.Sam.iii.5.2).
We arc rather fortunately placed as regards this legend, for the Rgveda has prescrved for us books of both familics. Both arc priests in thc scrvice of king Sudās, who could hiinsclf cxcrcisc pricstly functions, bcing the reputed author of x.133. The scnior pricst is Viávūmitra, thc cponym standing for the cntirc group ; thc gotra namc, as has been shown, is rcally kufika = the owl, a good bird totcm. A famous hymn is iii. 33, by Viśvāmitra to the two Punjab, rivers Vipas and Sutudru which lic crosses with hcavily loaded wagons of the Bharata tribc. This is apparently referred to in ii.53.9 and 11, where Sudās: is the king is made to cross safcly by Viśvāmitra, while iji.53.12 calls down a' blessing of Visvūmitra upon this tribc of the Bharatas. The implication is that Sudās and Viśvāmitra arc Bharatas. This sccms to be partially con-' firmed by vi.16.19, where the ancestralfire of the Bharatas is called the lord of Divodāsa, which is thc name of Sudās's father or paternal ancestor. .
V But thc Vasişthas also claim to bc thc pricsts of Sudás, in their own book, and there is amplc support for this. This disposes of thc fiction that the gotra of a ksatriya is that of his pricst, for it would follow that Sudās Paijavana changed gotras or had more than onc! We have to examinc the question of priority betwccn thcsc two clans which'occupy the pricsthood in succession for the same pcoplc. Hcrc for once we havc uncquivocal testimony : "Like sticks used to drive oxcn werc thc Bharatas split and cnfccblcd (= arbhakāsas ;
*The Pancavimsa Brahmana (xxi.12.9) may refer to some other Vasişthn in calling the sccr thc son of Vidu, though the same accounts calls Visvamitra king of the Jahnus, which would seem to refer to the two founder this.. But the conflicting double account of Vasistha's birth in RV. vii, 83.11-19 whcrcin hc is born of a water-goddess as well as from a jug which rescived the semen of Mitra-Varuna, will appear to be of special interest in tlic scqucl.

Page #16

36
D. D. KOSAMBI
according to Sāyana, "with few children'') ; then Vasiştha becamc thcir chicf pricst (purohita) and from the Titsus developed progeny (višas)" (vii,33,6). The statement is perfectly clear, and the special Vasiştha prayer for issue is to be scen in vii:4.7-8. Our versc abovc mcans that thc Trisus were a branch of thc Bhiaratas_though the name is taken by somc as synonymous for all thc Bharatas, which looks unlikcly unless it is from some other language. Vasiştha was not originally their pricst, but hc became thc purohita at somc later stage, and then the tribe multiplicd. Actually, in vii.33.10-11 Vasiştha derives his origin from Mitra-Varuna' and the very next versc from an asaras, both of which mystcrious legends have bccn amplificd later. This, with thc abscncc of an animal or trcc totcm, would strengthen thc implication that Vasiştha (whose name is merely an adjcctive proclaiming his superlative glory) was not as other Aryan mcn. On the other hand, hc cannot be taken as a divinc hcing
1 Of course, we have other descendants of the gods. Bhrgu is somctimes a descendant of Varuna: X.16-19 arc by sons or descendants of Yama, x.135 by a Kumara Yamayana: x 161 by Yami. . Arnong sons of Indra arc counted Virnada, author of x.20-20, the ape Vrsakapi of x.80, and Jaya, author of x.180. All gotra names ending in slamba are Bharadvajas,
1 Apsu is good Sumerian for the sweet (potalle) waters Ixoth alove and below the carth and the apsaras (in spite of Graumann's derivation at patas) is a water dcity. One may note other Sumcrian ciemenis in plenty. The gox Anu might even have been worshipped by the Anu tribe of the dásarajna enerics of Sudás. Certainly, Enki has features that remind us of Yama; his lucing a god of the absu, and slccping (after creation), are reminiscent of the later Indian Narayana. The reason for not diving way to this sort of speculation alsout the Slimcrians is the lack of clcar dorurnentary conncction between the two cultures, and the great difference in dates, though nothing prevents the legends and cults being common property of pre-Aryan pcoplcs some of whom later becamc Brahmins. The other difficulty is our ignorance of the actual phonetic values wliich were assigned 10 Sumerian ideo. gratis at various times and places; a personage whon textbooks of a geticration ago called the "batesi Of Shirourla'' is now known as the "isag of Lagash.” So, those who wish to follow in the footsteps of
* Waddell should find plenty of latitude of their conjectures. Let me present coniccture-nongers with the undeniable fact that the spotted cow l'Fani containing the sun, moon, and stars (viii.94,2) fits
an picture of the world.cow while Rgveda ii. 13.8 mcntions a Närmara whom someone is eure to eduatc to the Pharaoh Narmcr-Mcncs. Then take the Yaksus (vii. 18.15 as the tubos and so on.
1 Savana's gloss as well as the Brhaddcvata comment accm to take māna in vii. 33.13 as referring to Avastva'i birth from a jar, bregotten from the joint semen of Mitra-Varuna, llcre, it may be pointed
a stands in a special relationship to Varuna not only in his descent but in the intensity feeling of guilt, demonstrated in the four hymns vii. 80-89, In víi. 84.4, the ser asks what his sin Poi wishes to stril:chim down; in 5, he wishes for relcasc from come ancestral transgression
the next sk plcats action against the singer's will and the seventh or ctrayal : ava drugdhäni pitryā srja; the next sk plcada action against the ingree with
service or thic humblest sort : aram dāso na mihase karāni. In vii, 87 the tone of self-abnegation is not so proninetit, but the final versc again ycarns for blamelessness before Varung
for blamelessness before Varuna,. The lyricfest of the lot, vii.89 seems charged with this scrise of guilt acquired by unspecified but neer kratrah samalia dinata pralipam jagama Suve; mpī suksalra mīlaya, Nothing of this is
out Varuna as the first-born, just, benign, and peaceful god. seer of the Higveda, though they all honour Varuna as the first-horn, just, benina
t Vacistha was really guilty of having alsandoned the an. We might venture upon the interpretation that Vasistha was really guilty of having
the yaifia and worship of the war-gord Indra; cestral cult in favour of more paying practices, such as the yajfia and worshin or the walls
In this case, vii.88.4 would cven acquire a it is for this that his descendants had to ask forgiveness. In this case, vii.88.4 would special significance in its statement that Varuna had made a si of Vasistha. When and
have been carly to give ther. Vasig has their
When and where this supp
Cat. Brah. xii.6.141, 11.4.4.2.ctc.), Aryan traditional priority in Fajna ritual (Bhaddevalá v, 166-160; Sat, Brāh, xít.8.1941
mass and this has left its mark upon the tyvedic wanderings extended far beyond the upper Indus land-mass and this has left its mark in tradition älso, Vasistha speaks of bcing taken up on the ship of Varuna, and Kaksivän eine
o Phuivut being sayed at sca by the Násatyas, Aship with a hundred oars (1.116.6) and tric daaraway from the shore would hardly be expected on the upper Indus or any lake in the P... Quite incidentally, the father of Bhujyu is Tugra (1.110,3,117.4), which is also the name of a of Indra (vi.20,8; vi,20,4; x. 49.4).

Page #17

ORIGIN OF BRAIIMIN GOTRAS
bccausc lc is actually the pricst of a decaying clan, and vii.18, which describes the victorics of Sudis over many hostilc kings, cnds with a description of the gifts to Vasiştha ; thcsc gifts would bc uncalled for if some of the victorics were not duc to a Vasistha's incantations. Thc first battlc (vii.18.5-8) is on the Paruşņi, but there is at Icast onc other in vii.18.19, on thc Yamunā. This virtually spans the whole of greater Punjab, if thc Yamunā is to be understood as the modern river of that name (though it has been suggested that thc name, indicating mcrcly thc "twin river", inight again denotc thc Paruşņi; but x. 75.5 which has thc only rgvcdic mcntion of the Ganga sccms clcar for our intcrprctation). Now wc havc noted that thc gcncral movement is to thc cast, specifically proved in this casc by Patañjali's remark that the adjective "castcrn" for Bharatas is superfluous, as there aren't any Bharatas cxccpt in the cast : bharala-visesanam prig grahanam anarthakam, na hy aprāfico bharat (commenting on Piņ.2.4.66 ; latcr commentators takc Auddalaki as an example of a Bharata). Whcncc Visvamitra's passage of thc Bcas and thc Sutlej must bc an carlicr cvent, and the priority of Viúvamitra is therefore not in doubt.
Thc inversion consists in that Visvāmitra is made the upstart by later Brahmanical tradition in direct contradiction to thc clcar historical development.
If Vasistha and Visvamitra were both Brahmins as the term is understood by latcr writers, and thc Aryan priesthood confined to thc Brahimin caste, the logical development would have been thic adoption of Vasiştha into the Vibvāmitra or Kusika gotra. The story of Sunahácpa (Ait.Brāh.vii. 13-18 ; the names of thc thrcc brothers arc a suspicious scature) docs show such adoption, cvcn of onc chosen as sacrificial victim (cf. v.2.7 ;i.24.12-13). Indeed this adoption with thc changed name of *Devarāta is made responsiblc for the double marriage restrictions upon the Devarāta gotra though contrary to the acccpted results of adoption in tribal socicty. Even to this day. Brahmanical marriage restrictions arc circumvented by adoption into somc other gotra, which also forfcits inheritancc rights. But Vasiştha is cmphatically called thc first Brahmin priest, whence Brahminism is forcign to the original Aryan system. It susficcd, thercforc, that Vasiştha be adopted into the tribe, i not necessarily into the gens of the original tribal pricst, Viśvāmitra. It follows that Viśvāmitra, though a pricst, is originally not a Brahmin ; this is . attested by his title of rjarşi, applicd also to scvcral other ksatriya priests, as' for example the five (supposed) authors of i.100, tlic threc of x.179.
Whilc references to Sudās and his victories are scattered throughout the Rgvcda (though with highest frcqucncy by Vasiştha), the name Tștsu_occurs nowhere outside the seventh book. There is a faint possibility that the who of thc Tịtsu group (including ancestors of Sudās) was adopted into, and not
*Sunahsepa, son of Ajigarta, is the traditional author of 1.2 -30; tho RV. knows a Bharata Devavāta in iii. 28.2-3. The intended victim must have been a Jamadagni (clip.23).

Page #18

38.
.:!:D. D. KOSAMBI
a splinter of, the Bharatas; but there is no clan name now extant which can be derived from Tștsu. The adoption seems at least to have been that of Vasiştha and went to the extent of a common style in hairdressing ; vii.33 begins by describing the Vasişthas as dakṣiṇalas-kapardinas, with hairtwist on the right side, and kapardin is used only of the Tțlsus (vii.83.8) in describing human beings. The actual practice survived late, as we see from the appendix to the Gobhila Gșhya-sutra* : "The Vasişthas have a hairtwist (or braid) on the right, the Atreyas have three twists, the Angirasas five scalp-locks, the Bhřgus have completely shaven heads, and the others wcar a crest." This is to differentiate between gotra-groups, and "the others" here are the Viśyā. mitras and possibly the Kanvas, so far as the main Rgvedic families go.
THE DEATH OF A PRIEST: TvĀSTRA.,'
7. The rivalry between the Vißvāmitras and the upstart Vasişthas is plentifully attested in later tradition, while iii. 53.21-24 arc stanzas which still... pass as curses against the Vasişthas, so strong that were one of them to hear the particular verses, his head would split into a hundred pieces (they are still capable of giving anyone a headache !). On closer reading, these stanzas actually, do seem to be a mixture of curse and lament that the Bharatas are
beginning to prefer strangers to their own, the ass to the horse ; there.is no. | reason to doubt that they reflect the displacement of the Kuśikas by the Vasis
thas. We are told (Bịhaddevatā v. 112-120) that Viśvāmitra was deprived of his senses by Vasiştha and speech (vāk sasarparī) had to be supplied by Jama:: dagni. The brief hymn x.167 to Indra is given joint authorship of Viśvāmitra. and Jamadagni, which supports this close association. It follows that here Jamadagni is not on the same side as Vasiştha and their separate, rivalry is: attested by Tait.Sam.iii.1.7. ; v.4.11. Later tradition makes Jamadagni a sage at once hot tempered and forbearing ; capable of stopping the sun yet... killed unresisting by kşatriyas; in revenge his son Paraśurāma completely wipes . out: all kşatriyas from the face of the earth thrice seven times-though the Vedas have nothing of all this (Jāmadagnya being merely the supposed author, of x.110). This is one more of the inversions, with passage of time and rise of the Brahmins : it was the ksatriya who did the killing, and not conversely. In fact, even the Vasişthas are supposed not to have escaped unscathed, for: the Bșhaddevatā vi.28,33-4 reports "Now in the fifteenth and in the eighth
*For this stanza and a careful discussion of gotra-pravara exogamy as well as correspondence, between the traditional lists and the classification implicit in Pāṇini's derivations, see John Brough; "The Early History of the Gotras' in JRAS 1946, pp.32-45 : 1947.pp. 76-90. Though the learned. author's approach and point of view are entirely different from those adopted in this note, it is remark. able that he reaches the conclusion that at the time of composition of the Satapatha Brāhmaṇa, the entry of the Jāmadagnyas into the Brahmanical fold was (probably) still comparatively recent. My thanks are due to Prof. Brough for suggesting some corrections, though we still differ on the main ques : tion.

Page #19

ORIGIN OF BRAHMIN GOTRAS
89
(stanza) of the hymn (RV.vii. 104) the son of Varuņa (Vasistha); whilc as it wcre lamenting, his soul bcing ovcrwhelmed with pain and grics, utters a cursc. Vasiştha was at that time pained as his hundred sons had been slain by Sudāsa who, in conscqucncc of a cursc, had been transformed into a dcmon (raksas). Such is the sacred tradition." Again, thc Rgveda docs not rcport this but thc Tail. Sam.vii.4.7 docs ; such a tradition in the face of all the savour supposedly shown Vasiştha by Sudās cannot be devoid of truth. I suggest that some Vasişthas were so killed, pcrhaps some of thosc not rcgularly adopted into thc Trisus. Killing the pricst or his son is a fashion sct by Indra himsclf in bchcading Visvarüpa Tvāştra, whose thrcc heads lc (or his doublc Trita, ii.11.19; 8.8.8-9) struck off. This counts as a sin only in far later times, while we still have thc Tvästrcya gotra (GPN.156.18) among thc Jamadagnis. The thrcc hcads of Tväştra bccamc varictics of partridge (Bhaddevatā vi.151) and two of these bird totcms ccrtainly remain in the gotra lists, namcly Tittiri and Kapiñjala, though ncithcr is among the Jamadagnis proper. For that matter thc dcmon Rävaņa, thc warrior villain of the Rāmāyaṇa in later and morc castern lcgcnd, counts also as a Brahmin, and surprisingly cnough thc gotra is found in thc Vasiştha group (GPN. 113.11, 177. 22,177.1) though Vasişthia is traditionally the chict tcacher of Rāma ! Even the mild Atris did not cscapc as is scen by Saptavadhri's prayer for rclcasc from imprisonment (v.78.5-6) and by x.143.1-3, 1.117.3, X.39.9, perhaps rcferring to Atri's rclcase from a ficry pit.
Thc lasso as a weapon of war is uscd by the Sagartian contingent of Xcrxes's cavalry (Hcrodotos vii.84), and by individual hcrocs in thic Shah Nameh. This may bc the original pāśa from which freedom is desired, pcrhaps symbolically, in several lymns. Thc gloss ascribes viii. 67 to fishes caught in a nct and praying for freedom, which could have been dismissed as a myth had it not bcen for the fact that the Matsya tribe appcars in vii.18; and in the Mahābhārata as the people of king Virāța. Thc Vaphio gold cups show us nets bcing used to catch wild bulls while the god Ningirsu is shown on Eannantums' stcla (stelc des vautours, in the Louvrc) cnfolding thc men of Umma in a nct and crushing those who try to cscape, whence its use for prisoners of war is also possiblc.
1 The burning by thc Saudūsas of a son of Vasiştha pamcd Sakti, is also reported by the Satyāyana and Jaiminiya Brāhmaṇas (H. Oertel, JOAS xviii, 1897 pp. 16-18, particularly 1.47). For thic cannibal Saudása in later fable, Jatakamäln 31 ctc.
Wc sccm to have a reference to both divinc and human imprisonment (of Brahmins) in iv, 12.5 ūrvād devānam ula mariysinfim. The grādiah pasuh (viii, 1.31) could be a Yadu prisoner of war, particularly when read with viii, 6.-18: Sravasi vidram janam. The traditional Yadu capital Dvāraká cannot have been the modern port of that name in Saurastra. We have a clcar narrative of the Yadus including Krana and Balarama, being driven out of Mathura by Jarāsamdha. They go westwards to found the new city in the safety of a mountain barrier : Mbh. 2, 12,9; 2, 13, 44, 49, 65. This is the logical dircction, considering the desert to the southwest of Mathura ; the original Dvärakā may thus have licen Darwaz in Afghanistan, or the capital of Kambojn in Buddhist records.

Page #20

GOTRAS
.
. ORIGIN OF BRAHMIN, GOTRAS ..
i .
41
4.1
..
.
duism never developed an established church, and that the Brahmin caste · began to serve the general population by ritual, rather than the warrior class by yajña, only after the rise of Buddhism. In the earliest days (as in Rome and Greece), it was the right as well as the duty of every head of a patriarchal family to perform priestly functions later reserved for Brahmins ; and knowledge of vedic Sanskrit was common without the prolonged study it necessitated later. If, under such circumstances, we find the beginning at least of endogamous castes, it is necessary to inquire what external forces would lead to and accelerate this type of partition. The major feature is the conquest; it will be shown that this does account for the Šūdra caste. But it is difficult to believe that no other. portion of the conquered population survived besides the helots; that we should nevertheless find the reappearance of Indus Valley motifs, including multipleheaded and many-armed deities--particularly Brahma. That cities like Mohenjo-daro could exist without class. divisions.is quite incredible in view of what is known of ancient society, and if their armament does not appear from known excavations (which are certainly incomplete) to have been very good, it implies the existence of some other method than pure force for maintaining the class division. This method," so far as known, can only be religion, and that in turn implies the existence of a strong, fully-developed, and well-organizcd priesthood. I may point out in this connection the importance of the desert bordering the river (as in Egypt and Mesopotamia) for this not only makes the development of agriculture, and later of the city-state, possible as well as necessary, but also economizes the energy spent upon defence against wild animals, barbarians, and in cutting down forests. The intervening desert is an excellent natural barrier against external enemies till they learn the advanced military technique necessary for crossing it and taking walled cities. The need for internal force is minimized by the priesthood. After the Aryan conquest, nothing would be casier than the absorption of some upper layers of the conquered society, and the most attractive would be the priesthood, even more important than the technicians in any primitive society. Of course, this would greatly intensify the development of classes aniong thc conquerors as soon as they began to settle down ; which is precisely what we find on comparing the Rgveda with the Taittiriya Samhitā and later documents. As further support, I might point out that a considerable num stories appear rather late, albeit with claim to antiquity-as for example the flood legends and the purūnas in general, though some of the material is undoubtedly pre-Aryan.
. In this direction, it is also necessary to remark that matriarchy survives only among the least Aryanized of the people found in India today. If the conqucred had even a remnant of this system, it would be casy for them to preserve their group structure for a while after adoption into various patriarchal gentes. Thus ive should not be surprised at finding Dirghatamas called

Page #21

ORIGIN OF BRAHMIN. GOTRAS
duism never developed an established church, and that the Brahmin caste began to serve the general population by ritual, rather than the warrior class by yajña, only after the rise of Buddhism. In the carliest days (as in Rome and Greece), it was the right as well as the duty of every head of a patriarchal family to perform priestly functions later reserved for Brahmins; and knowledge of vedic Sanskrit was common without the prolonged study it necessitated later. If, under such circumstances, we find the beginning at least of endogamous castes, it is necessary to inquire what external forces would lead to and accelerate this type of partition. The major feature is the conquest; it will be shown that this docs account for the fudra caste. But it is difficult to believe that no other portion of the conquered population survived besides the helots; that we should nevertheless find the reappearance of Indus Valley motifs, including multiplcheaded and many-armed deities-particularly Brahma. That cities like Mohenjo-daro could exist without class divisions is quite incredible in view of what is known of ancient society, and if their armament does not appear from known excavations (which are certainly incomplete) to have been very good, it implies the existence of some other method than pure force for maintaining the class division. This method, so far as known, can only be religion, and that in turn implies the existence of a strong, fully-developed, and well-organized priesthood. I may point out in this connection the importance of the desert bordering the river (as in Egypt and Mesopotamia) for this not only makes the development of agriculture, and later of the city-state, possible as well as necessary, but also cconomizes the energy spent upon defence against wild animals, barbarians, and in cutting down forests. The intervening desert is an excellent natural barrier against external enemies till they learn the advanced military technique necessary for crossing it and taking walled cities. The need for internal force is minimized by the priesthood. After the Aryan conquest, nothing would be casier than the absorption of some upper layers of the conquered society, and the most attractive would be the priesthood, even more important than the technicians in any primitive society. Of course, this would greatly intensify the development of classes aniong the conquerors as soon as they began to settle down; which is precisely what we find on comparing the Rgveda with the Taittiriya Samhita and later documents. As further support, I might point out that a considerable number of ancient storics appear rather late, albeit with claim to antiquity-as for example the flood legends and the puranas in general, though some of the material is undoubtedly pre-Aryan.
41
In this direction, it is also necessary to remark that matriarchy survives only among the least Aryanized of the people found in India today. If the conquered had even a remnant of this system, it would be casy for them to preserve their group structure for a while after adoption into various patriarchal gentes. Thus we should not be surprised at finding Dirghatamas called
6

Page #22

42
D. D. KOSAMBI
Mămatcya after his inother, a custom to be observed in the final Bphadāranyaka Upanişad linc of teachers.
Onc sign of conflict between the Bralunin and ksatriya castes, after full dcvclopment of the system, appears in the original mcaning of x.109, which sccms to have been composed for the return of a Bralımin's wisc abducted by a ksatriya. Onc obvious reason for the later appcarance of the Jamadagnis and the still later risc to pre-eminence of Blițgu is this previous enmity. Thesc people were still being killed by thc ksatriyas when the Viśvāmitras werc bcing ousted by the Vasişthas from thc Bharatan pricsthood. The objcction will undoubtedly be madc that thc later Brahimins could have been Aryans from some extra-vedic branch. Why could thc Jamadagnis, with their Indo-Europcan namc, not have been vrátyas ? In the first place, thc vrătyas were first differentiated from the rest long after the Bhrgu-Jamadagni group was well established (though not necessarily in all parts of the country) and the vedas fully developed. In thc Rgvedic age, thic term vrátya could not have been used to distinguish extra-vedic Aryans bccausc all Aryans were then wanderers while the development of the veclas itself reflects the risc of settlements. The vrátya tribes do not need the vedas simply because they continue to wander castwards, into territory without a great civilization comparable with that of the Indus vallcy. At that later stage when the vrätyas proper have to be distinguished, the adoption of their pricsts would not only be unnecessary but highly improbable for the simple reason that their priesthood --if indeed it had a separatc cxistence-would be much less developed than that of the main vcdic Aryans. This can be seen from the vratyastoma ritual, created later for the adoption of a vrătya, not of his priest, into the vedic fold; from book xy of the Atharva-veda, which seems written to placate the vrátyas ; from the term brahmabandhu, applied to Magadhan Brahmins who associated themselves with the vrătya ceremonies, and even now used of Brahmins without Icarning. The great vrâtya tribc is that of thc Licchavis, mentioned with respect by the Jains, and the earliest Buddhists, while maintaining a high social position down to the Gupta period at least. We have yet to hcar anything of their priesthood. The philological argument from the name carries less force now that Hittite records have been read ; also, adoption being a form of rebirth, a non-Aryan name would be thc first to change. Even without adop
* Sarabha is called rşibandhu in viii. 100.6 but without the forceful contcmpt that goes with the termination bandhu later on. The Licchavis arc ksatriya vrătyas according to Manusmrti 10.21 var. nicchiri), known to Buddhist litcraturc gcncrally as Vajjis (=thc wanderers). Patanjali on Pan. 5.2.21 : nānājātiya aniyalarllaya ulscdhajirinah samgha urītāh shows that any tribal organization outside the Brahmin ritual and four-caste system could be called vrătya, forcshadowing modern guild-castcs and professional tribes, Thc Mahāparinibbānasutta shows that the basic rules of the Buddhist samgha were derived from Aryan tribal constitutions, specifically that of the Licchavis. For a survey of the Brahmin literature (without rcalization that sutras concerned on y, with reconciling vrätya ouservances with vedic ritual say nothing about the actual lifc lcd by thc tri hes) Scc J. W. Hauer, Der Vrdlya (Stutt. gart 1927 ; vol. I only). The vrätya Gyhapati of Pañc. Brah. xvii, 1. 14, 17 could easily be the tribal chief with the usual priestly functions, and no othcr explanation will fit as well.

Page #23

ORIGIN OF BRAIMIN GOTRAS
43
tion, conquest favours a new stylc in names, as seen from Greck names adopted by Jews. Proponents of the "extra-vcdic Aryans" thcory would have to postulatc rathcr complicated rclationships bctwccn groups of Aryans not known to have cxistcd ; at the same timc, thc rccurrcncc of Indus Valley types in later iconography would be very difficult to cxplain.
ADOPTED PRIESTS
8. Looking closcly at the first list of Sudās's cncmics in vii. 18.5-7, we find the following* : Simyu, Turvasa, Yakşu, Matsya, Druhyu, Bhrgu, Paktha, Bhalāna, Vişñņin, Alina (and perhaps Thic Sivas) ; in vii.83.7, the ten kings opposing Sudis arc called ayajyava), "without the fire sacrifice." The notable occurrence here is of the Bhrgus, who cannot then have been merely Brahmin priests. This is to some extent supported by the surviving designation bhārgava (?“tlic roaster") for a potter, which is quite natural if fire were the particular technique of the Bhrgus, as it appears to be in the Rgveda. Their chariot receives special mention in iv.16.20 and x.39.14 by the phrasc bhrgavo na ratham. Hence, they arc a complete tribe, with all the professions. If their name survives only as that of a Brahmin gotra, it must be because some of them managed to become pricsts of the Aryans. That they were not always Aryans themselves would follow from vi.18.7, which specifically mentions Indra, as the friend of the Aryans, bringing aid to the other side. That the Indus vallcy culture could exist without strong class differences is incrediblc, and their pricst class must have had specially refined ceremonial, which would cnablc them to be adopted fairly casily into the priesthood of the conqucrors, provided they took up the new cults.
It is quite simple now to show that there arc other elements besides the Bhỉgus which are so assimilated. Kutsa, for example, counts as a Bharadvāja gotra with Kautsa (GPN.63.14,165.21,61.4); it is therefore natural to find Kutsa the author of i.94-8. But in the body of the vcda we read conflicting stories about him, for hc is at times favoured by Indra and at times crushed ; thc first may be seen in x.49.4 and the second in i.53.10. This can be explained
*Some of thesc names may be adjectives, but this scems a reasonable way to make up the dāšarājfia ; ¡ust which people arc qualified as Šivāsas is not clcar, so possibly The Sivas arc one more tribe. The econd battle is assumed to be distinct from the first.

Page #24

1). D. KOSAMBI
by our present thesis of progressivc assimilation of a Kutsa tribc. The l'arus* arc mcntioned in i. 108 with the Yadus, Turvalas, Druhyus, Anus, all originally liostilc to Indra and the Aryans. The particular portion of the hymn is uit. doubtedly latc; but it is to be note that Yadu, Turvasn, Anu, Druhyu, Pūru arc all five made sons of king Yayati (the fir:llwo by Devayani, a Bralimin
(Bhrgu) daughter of Sukra, prcccptor to the Asuras) in later lcgend (Mbh. 1 1.78.9-10). Krkņa (=black), the incarnatc god of the Mahābhārata, is himself : a Yadu. A Kánva named Krşņa is thc poct of viii.85, a hymn addressed to the
Asvins. It is notable that the scer is called "black" by name, like thc Angiras author of x.42-44 ; in the Atharva-vcda ii. 25, kanuc mcans "cvil spirit”, to be is exorcised. It would be simplest to regard this not as a fortuitous coincidence, . but as indicative of some Kanvas liaving heen adopted from the dark prc-Aryans,
of whom thc unadoptcd portion was submitted to the usual process of dcmonization with the passage of time. Just when thicsc fivc pcoplc bccamc Aryans is not
clear, but certainly the bravc king Pūros dcfcated by Alexander in his invasion of “the Indus vallcy is (with his nephicw) the last Paurava known to history, so that
somcöf thesc ancient lincages actually cxisted down to a latc period, and had to be cxplained by a suitably rewritten tradition. This tradition never disguises the hostility between thc dark (hcncc un-Aryan) Krena and Indra, which seems to go back to viii. 96. 14,19 (accepting the reasonablc Sāyana gloss). Wccvcn get the Purukitsa combination as a king-namc, probably thc rcpresentativc of an amalgamatcd tribe; in our Brahmin gotra lists the name is found among thc Bharadvājas (GPN.61.14), which would be impossible cxccpt on our hypothesis. In fact, references in book vi. make it clear that some Bharadvājas were pricsts of Purukutsa's son, hence the formation of that gotra among the Bharadvajas. The descendant Kurusravana cmbodics the first mention of thc Kurus, in his name (x.32.9;33.4).
'When we come to kevala groups, thc origin of the inverted rule that the · priest's cotra is that of the king becomes still clcarcr. Vitahavya is a Bhārgaya
Golra (GPN.34.4-5) but thc Ssõjaya Vaitahavyas of Atharva-vcda v.18-19 arc rude kaatriyas who slaughter Bhrgu's cow ; thc sage is lielpless and the cow herself takes revenge upon the insolent warriors, who are destroyed. But this
Identification is particularly difficult in the care of the Purus, for the narar for plenitude, or a Irine of people in general, as well as a specific tribe named For Town the
articular tribe is meant, lcing mentioned along with Tyksi and the Druhyus: similaritin g hunt their notition even as against thic Vasişthas and suns 19 not clcar, for vii. 18.13 aprake it of having bcaten or of hoping to defcat (jesma) the tricky-voiced Puru" in vit 10 of as having helped the l'uru kiny Trasad?syu ; in vii: 96.2 the phta . stood by come as indicating that the Purut vcrc then criticd on both hanteca? that the Purus vcrc then critid on loth banks of the Sarasvati. The adjcc
Praca nas been minalo vacū of thr: Panisin vii 0.%. thc Dasyus in xv.26.10. Sudás, might cven be made out a Puruly i.03.7. Hopkins in JAOS xv. pp. 262-283 outdocs the most ridicu dits in deducing that the datarajia was conspiracy. Icd or instigated by Vice
he his main discovery, namely that vii...contains dcrisive allusions to Visvamitra as often as possiblc. Just how this escaped the Indian tradition, which is sure
* vet to be explained; but undantly rome linca) dcrcendant will appear to reach ornicion made by Hopkins !

Page #25

ORIGIN OF BRAHMIN GOTRAS
46
would Icavc thc gotra and pravara uncxplained, so we have a still later story to round out the narrative, that Bhrgu magnanimously and magically converted
thc rcfugcc Vitahavya into a Brahimin mcrcly by telling his pursucrs that there : was no ksatriya in thic hermitage. Vitahavya as an adjective is applicd to
Sudās in vii. 18.3, presumably in thc scnsc of hc whosc libations arc agrccablc to the gods; thc namc occurs as that of the author of vi.15, and cxplioitly in vi. 15.3 calling down blessings upon him. But thc sixth book is of the Bharadvājas, whicncc wc again have a contradiction. This may be resolved by the explanation that some Sșñjaya Vaitahavyas, not necessarily connected with the singer of vi. 15, had a Bhrgu as thcir family pricst. But inasmuch as the kşatriya was not by any means cxcluded from the pricsthood, properly the function of any, tribal lcader or family head, those Blirguids who survived in this particular linc had to bc adopted by tribal rulc into thc Vitahavyas, whence by the later antithctic inversion we get the formation of a Vitahavya pravara among the Bhrgus. This process is very clear among the ten cxtra familics cnrolled among thc Bhřgus and Augirasas, as shown by the king-names that form the supposedly Brahmin pravaras. It will be fairly obvious that, at Icast as regards thcsc spccial. kevala families, the pravara develops by adoption by some Brahmin group of a kşatriya family namc. Mudgala is a Kevala. Augiras group in the lists, but the Mudgala of x.102 is a splendid fighter. Though not in thc Vcda, the Purāņas makc Vişņuvěddha son of Trasadasyu, hcncc a kşatriya, though the name is in the Brahmin pravara lists. Thc Vena of x. 93.14 sccms to be a non-Aryan king.
With the exception of people within the tribe or cult, as in the case of Indra himselfor Sudās, Rgvedic names of a tribc and its leader seem to be identical, particularly in speaking of people not intimately known to thc hymnsinger. This agrees with what we know of tribal socicty in other places. The MaoDonald would be thc head of the clan MacDonald in Scotland. Not less than ten different lcadcrs named Appius Claudius hicaded the Claudian gens in Romc after its incorporation under the first Sabinc head (Attius Clausus) ; if Latin records were as diffusc as the Sanskrit, the deeds of all of them would havc been incxtricably confounded. The distinction between hсads of families and ordinary members appears in Pāņini's grammar (sce J. Brough, loc.cit., for thc significance of the yuvan). Vd. xii.7 shows that the period of mourning - among the Iranians for the hcad of a family was six times that even for a parent. For my purpose, the designation of wholc clans by a single rşi's namc (for onc
vimitra or Vasistha can hardly have composcd the entire books in thcir respective names) yields further support for the adoption theory. What nceds carcful proof is the statement that some of these adopted priests must have belonged to prc-Aryan Indian groups.
The distinction between Brahmin and ksatriya priestly traditions even after their merger may be seen in the position of Vişnu, who is a very minor
.

Page #26

46
1
god' in the Visvamitra book.1 But three complete hymns to Visnu by Dirghatamas (i. 154-6) show a totally changed relative emphasis. The sage himself, according to the Bṛhaddevatā iv. 11-30, was the blind son of a Bhṛgu mother, and in his old age cast into the river which carried him safely eastwards beyond the Aryan pale to Angā.
PRE-ARYANS AND ARCHAEOLOGY
9. It is still necessary to show that some of these new recruits to the vedic fold were non-Aryans,2 for there is no doubt that there did exist non-vedic Aryans; among the Indians, it sufficed to refer to the vratya Licchavis. So, it might be suggested that the whole fight with Sudas's enemies was in fact a
1Apart from stylistic and metrical variation, as well as the subject matter, the mere incidence of hymn dedications my be used as a guide to clan differences. The ninth book, being dedicated to Soma, and the Välakhilyas as later appendages, have been discarded; in doubtful cases, I follow Grassmann as far as possible. The standard hymn order within books or groups allows us to emphasize dedications to Agni, Indra, and all the rest. Among the "rest" have been counted even those hymns where Indra or Agni, or both a have a share. This gives us the following table:
Book
ii
iii
iv
>7 ×
i
Agni
10
D. D. KOSAMBI
29
15
28
16
17
14
45
30
204
Indra
12
24
17
12.
31
15
45
41
44
Rest 21
9
26
47
28
72
33
105
117
458
Total 43
62
58
Totals
241
903
Modern statistical tests give information that agrees very well with what we know from other considerations. The Visvamitra book (iii) differs from all the rest, as would be expected from the real Aryan kṣatriya tradition. Books i and x may be grouped together. Books ii.iv., vi can also be combined among themselves, which proves the Bhrgu-Angiras unity of dedication. The Kanvas are closest to this group in spite of their great predilection for Indra, while only Atri comes near Vasistha, though none too close. (Calculations by Mr. S. Raghavachari for the chi-square test). In support, we may recall that the eighth book, though Kanva by tradition and with a good unity of metre and style, is unquestionably of mixed authorship; not only other Angirasas but Atris, Bhrgus (including Jamadagni and Uśanas), Kasyapa, possibly a Vasistha Dyumnika (viii.87), Trita Apiya (viii, 47, but this is impossible as the final verses show), and even Manu Vaivasvata are given a share in the authorship, by the Anukramani tradition. Only Visvamitra is stubbornly excluded, and this is highly suggestive.
87
75
104
92
191
191
* Traditionally, the Soma book contains eight hymns ascribed to a Kavi Bhargava, who is identical with or the father of Kavya Usanas, who is in turn the author of three more. But the famous Devayani story of the Mahabharata shows this personage as preceptor to the Asuras, which can be explained only on our present hypothesis of assimilation of non-Aryan priests, not necessarily in India. In the Rgveda, Uśanas is mentioned almost exclusively by the Angirasas: i.51.10-11 (Savya); 1.83.5 (Gotama Rahugana); i.121.12 (Kaksivän); iv.26.1 (Vämadeva); vi.20.11 (Bharadvaja) ; .iii.23.17 (Viśvamanas, son of Vyaśva); ix.87.3 (Usanas himself!); ix.97.3 (Vrsagana, supposedly a Vasistha) ; x.40.7 (Ghoṣā, daughter of Kakṣivän). Otherwise usana is desire, of which Grassmann takes the name as a masculine personification. One can't expect this in Angiras books, where Bṛhaspati is an Angiras (vi.73.1) and even Agni (viii.84.4) in a hymn ascribed to Uśanas. Without discussing his dentity with Kai Kāōs or Kavi Usa of the Iranians, it is fairly clear that he must be a figure of the iransitional period.

Page #27

ORIGIN OR BRAHMIN GOTRAS
4.7
civil war among Aryans (as in partit must have bccn), that the hostility which can so amply bc proved down to later times is professional, between thc warrior and pricstly castcs, and at most derives from thc ancient hostility among diffcrcnt Aryan tribcs. After all, Kurunga is called a Turvaga king in his dānastuti hy Kaņva, viii.4.19, and could be an Aryan; morc ambiguously, Kanva bogs Indra to let him scc Yadu-Turvaša again in sk 7. Not only in vii.83.1, but also in other hymns (vi.33.3, vi.22.10, vi.60.6) arc both Aryan and non-Aryan (called Vrtras hcrc) cncmics mcntioned whicn praying to vcd tcction. In iv.30.7, Indra takes Yadu and Turvasa across dry (or unbathed ; thic mcaning is obscurc) but kills two (presumably non-vcdic) Aryans Arņa and Citraratla on thc other side. There is, then, cvidcncc for the progressive rccombination of Aryans and non-Aryans into vcdic and cxtra-vcdic group. In vii.83.1, Indra-Varuņa arc to stand by Sudās and strikc cncmics, both Aryans and Vștras. But our point is casily proved.
Tura Kāvaşcya is famous teacher in the Satapatha Brāhmaṇa, a lcading pricst in thc Aitarcya Brāhmana, and prominent in other Brahimin tradition, though thc direct gotra docs not. sccm to lavc survived. But his father, Kavaşa Ailūşa (sccr of x.30-34 and pricst of Kurusravaņa, x.30) is forcibly cjccted as dāsyah pulral* by Brahmins, to dic of thirst from which hc is saved by his river hymn (x.30, because of which the sacrcd river Sarasvati followed him into the dcscrt ; cf. Ait.Brāli.ii.19). Thc ancestral representativc Kavaşa is overthrown in vii.18.12 along with thc Druhyus, which should complctc thc story. But it might still bc objccted that dāsī mcans only a slave girl, and there is nothing to show diffcrcncc of race, cvcn though a slave girl's son would certainly be disqualificd.
Däsa in the grcatcr number of rgvcclic citations mcans a human foe .conquered by thc Aryans in battic, Indra yathāvašam nayati dāsam äryaḥ (v.34.6 and others). They havc thcir own citics, strong.cnough to be called brazen or iron; (ii.20.8) hat dasyūn pura āyasīr ni tārīt. Dasyu is taken as synonymous at times with dcmons, and again with dāsa, which shows that the strisc is very old : (iii.12.6) Indrāgni navatim puro dāsapatnir adhūnutam. Some of these citics arc scasonal, particularly autumnal (sāradīḥ) : sapta yat puraḥ sarma säradar dard han dāsih Purukutsāya sikşan (vi.20.10) which incidentally show that Purukutsa was befriended by Indra at that time, whatever thc components of his name may have been carlicr. Thcy have a special colour yo dāsam varnam (ii.12.4) which is not that of thc 'Aryans: hatūí dasyūn prāryam varnam ävat (iii.34.9). They! arc always different in religion (cf. Manusmfti x.44,45), which is of far grcater importance than the colour. They havc not the firc-sacrificc : ayajvānaḥ (i.33.4), nor thc proper cult and arc possessed of black magic : mājāvān atrahma
* A similar rcproach by Medhatithi against Vatsa Kāņva - was disproved by the accused (Pañc. Drah. xiv, 0.6).

Page #28

48
D). D. KOSAHPI
dasyur arla ( iv.16.9), besides being black and posecesed of citics : (iv.16.13; pañcīšat kısņā ni vapaḥ sahasra alkam na puro jarimā vi darda). They arc trcacherous, without the Aryar. observances, and hardly human (2.22.8) : akarma dasjur abhi no amantür anyavrato amănunah ; lvam las; ūmilrahan radhar dāsassa dambhaja. Just what the designation anāsas ("noscless" or "faceless" ?) applica to them in v.29.10 mcans is not cicar, but it surcly refers to thcir different appearance.
/
Only in three cases docs dūsa clearly meani a servant or slave, an carly reference to the hclotage to which a great part of the subjected people sank.. Of thesc, 7.62.10 referring to gifts made by Yadu and Turva to the hard mcntions cither two slaves or portrays the humility of the donors, but the names as well as the reference being part of the dānastvli may indicato a latcr addition. The reference viii.56.3 in a Válalhilya can be ignored. In vii.85.7, the scer spcaks of scrving the god likc a disa, which can only can slavc or servant, not cnemy. The rare mention shows that the new relation was crncogent, not fully established. Therefore, we are lcd to wonder whether Divodása mcans "slavc of heaven", or whether the period is carly cnough for the name to indicate a disa viho had been adopted by the other side. I myself incline to the latter interpretation, sceing that daca has gencrally thc mcaning of a specific people from whom thc kūdra castc and scrvitudc dcscloped by conauest. In any case, the termination dása as part of a name is not to be seen clscvihere than with Divoclása land in later orthography luis son “Paijavana" Sudāj. Trasadasyu (çon of Purukut:a) do not scom to mean thc Dasyu namca Traca, but he who makes thc Dasyus tremllc.
The concept of tribal property in a migratory pastoral socicty enables us to sketch an outline of development for the südra caute. The Indus valley citu dwellers could not have been fod without a comparatively large ancillary orarian population. The invaders' way of life made such prisoners us:1033
at first, for without agricultura a human being could produce very little surplus • beyond that necord for his own maintenance. A prisoner would be sacrificed or adopted, as the Sunahkcpa story tcil: 115. If the agrarian population of the
i had been clicctive as fighters, the conquest would not have taken
straut not becii so devastating. They must have been ton numcrous at en masse, but not dangerous enough to be killed off alvotcther. Thus isors would form a group by themselves and whatever the could w their own methods, as well as their labour, would belong to the
may whole. This is the first castc, initially a difference beand dágas, as the word rama for Castc and colour shows in con
berardod fact that thic Aryanliarla diffcrc:it colour from their predecessors in India. However, the existence of a castc, of urplus
id class and costo differentiation among Jabour, would necessarily promote rapid class and castro cliffor.
placc, or at Icast not hoci so devast
conqucring tribc as a whole. This is lwoon Aryans and dásas, as the word van junction with the recorded fact that the Aker

Page #29

ORIGIN OF BRAIIMIN GOTRAS
49
the conqucrors; it ccrtainly inhibited thc risc of largc-scalo chattcl-slavcry v in India. The wandering vrâtyas alonc prcscrved the older tribal institutions down into historical limes, having need of neither Brahmin nor dūdra within the tribc; kingship with thcm remains a tribal olicc of small importance cvcri whicn the tribc developed into an oligarchy over a tributary population. Bralımin scriptures continue to give a fixed thcorctical status of a hclot for thic sūdra, always distinguishıcd from Arya : one who is not cligiblc for initiation, barred from handling wcapons, cvcn from owning properly, and whosc function is solcly to labour for the bercfit of the thrcc higher castcs-though we know that this was on occasion tacitly contradicted by the risc of a sūdra to the position of warrior, chicstain, or king, in historical times
Thc conqucrors must have destroyed cultivation as well as thc citics; otherwisc thcy could havc scttled down likc thcir cousins in Egypt, Asia Minor; Mesopotamia as a new layer on top of cxisting class-relations of productiori. It is well-known that without irrigation thc Punjab plains can support only a comparatively small population along the rivers, thc rcst being desert. Nowhcrc in Alcxander's timc do wc hear of any citics comparablc in sizc and organization to Mohenjo-daro. On the other hand, we find the common vcdic myth of Indra killing a dcmon to free the pent-up watcrs (sometimes called cows), which is invariably taken to dcnotc a naturc myth of the rain-god picrcing clouds to cause precipitation. But we have a scparatc rain-god parjanya to whom cntirc hymns arc dedicated (v.83 ; vii.101,102). Indra's action is described in terms that can only mcan that the river-dams were shattered; wc know that a littic to the west of Mohenjo-daro, thcrc still exist tremendous : prchistoric dams of this sort, though now uscless in the absence of sufficient rain (Marshall, p.3). Thc brcaking of dams would destroy thc very basis of agriculturc, whcncc thc Aryans would have to move their cattlc to fresh pastures after a few years. Perhaps the clcarcst dcscription is to be seen in iv. 19.5,4,8:. Indra shakes the ground as the wind thc watcr, overthrows the mountains, forcibly bends down what was firm ; the rivers hastcn forth, all the stones roll away likc chariots ; for many days and.ycars did Indra lct the rivers run after the fall of Vộtra, hc freed the strcams that had been bound (badbadhanah sītäh, thc dammcd rivers*). Only ignorancc of the fact that therc had been a civilization with fully developed agriculture in the desert, before thc Aryans, could make anyonc intcrprct this as a myth of rain-making. Similarly for i.32.8-10; viii.96.18; wc hcar of scasonal barricrs in v.32.2, and vii.18.8 spcaks of vain attempts at divcrting thc Paruşņi river, perhaps one of the causes of Sudās's wars. In ii.15.3 vajrena khāny atšņan nadīnām has been interpreted as Indra making canals for the rivers, but this quite uniquc action on the part of
*The particular word for dam might have been rodhas or rodhana, i.38 11 ; ii. 13.10; iv. 22.4 ; x. 48.2. In ii, 15.8, rinas rodhamsi krirināni shows that the obstaclc removed by Indra was artificial, not natural; the other references can at worst be taken to mcan walls or river-banks,

Page #30

50
· D. D. KOSAMBI
· Indra may be doubted, becausc thc verb and tool both indicate, smashing, which is possible for a dam, not for irrigation channels. Besides the dāsa as a source of labour power, the humped Indus cattle were also an acquisition of thc.conquest; they are mentioned cxplicitly in x.8.2: x.102.7. and perhaps in viü.20.21, and their truly Indian origin has generally been admitted. Thc usc of the horsc and of iron was known to the invaders bcforc thcir, irruption, according to archacologists. We have here onc rcason for the victory of the Aryans.over the indigenous population which know ncither... ....
.:. Heterogencity in the pre-Aryan pcople cannot bc doubted. They cannot all have been residents of Mohenjo Daro and Harappa with but a single profession! Even to support thc inhabitants of a big city likc cither of thesc, there must have been, a considerablc food-growing ancillary population apart from the craftsmėn in the citics, of whose social position wc still know nothing, but; who would be the logical candidates for the name and position of Disa, or Dasyu. All I can suggest is that a portion of the conquered rosc instead of falling, and that they could only do this by adopting Aryan methods of fire worship, undoubtedly with some additions. Some of them must have had fire cults of thcir own, as for example the Bhrgu-Augiras group so often associated with the first discovery of firc. A few like Divodása* may cven have been enrolled into the ruling kpatriya class, for the Aryans had come across many different, pcople in their wanderings, and purity of "race" at so carly a period means nothing in comparison to the cult obscrved ; adoption of a stranger needs only the formalities of initiation, and one becomes a ksatriya mcrcly on account of prowess in battle. It seems clear to me that the formation of an internal,. Aryan.caste, system, csscntially the scparation of the Brahmin in function and discipline, from the kgatriya and the setting of both above the houscholder:vaisya, after the dāsas had been conqucrcd, must have been acccIcrated by, thc assimilation of:a subjugated priesthood; for otherwise there is no reason for;demarcation into endogamous castes. The Indic Aryans com oleted their own conquest at a far earlicr pcriod than the Zoroastrians (identifving Vistāsp with the father of Darius. 1, after Herzfeld p.30, p.88) with more primitive tools and over, cultures which were far more locally concentrated. The question can only be settled, with more archacological evidence; the purpose of such a discussion as, the present is primarily to show the intelligent archacologist what to expect;, perhaps where to excavatc, and how to interpret his finds.:,:::... .:::: : . . ...... ........! !! ': As a preliminary,.connections may be pointed out between certain obscurc featuresof, the, Rgveda; and actual finds in the Indus vallcy... The threc
Saraswau to Vadhryasva, according to vi.61.1. As Sudás is both - Divadisa is a gift of the river Sarasvati to Vadhryasva, according to win Daliavana (though no Pijavana is known) and a descendant or son of Divodása, there is som of adoption herc... ,..
or son of Divodása, there is some possibility

Page #31

ORIGIN OF BRAHMIN GOTRAS
3
...
headed scated deity of the famous Mohenjo-Daro seal, our fig.1; may be taken for: Tvastra, if the nuinber of heads be actually three; there may be a fourth head away from the observer, which would make the deity proto-Brahmu.' But the three-headed Tvästra cannot be entirely independent of other threeheaded creatures on Indus valley scals. In E.J.H. Mackay's "Further exca vations at Mohenjo-Daro" II, Pl. LXXXIII.24, XCVI.494, XCIX.B and Marshall's carlier work (Mohenjo-Daro and the Indus civilization, London' 1931) III. Pl. CXII. 382, we find a scal depicting a three-headed bull. Now iii.56.3 refers specifically to such a bull in the Rgveda; while the entire hymn is to several otherwise mysterious multiple deities. So far, it has not been possible to demonstrate dakṣihatas-kapardinas on any scal but a god 'with braided hair is to be seen in our fig.2 (Mackay Pl. LXXXVII. 235); the god, along with a priest and a row of seven human figures who are attendants at the sacrifice all show long hair-braids (in Mackay II; Pl.XCIV. 430, PI.XCIX-A Marshall I, Pl. XII.18). Kapardin should rather mean with twisted than braided hair, but the matter is not settled. Punch-marked coins,also, yield occasional homo-signs with hair-twists or braids (Durga Prasad, JRASB.XXX.1934, Pl.21, nos. 132-3) but the coins belong to the Mauryan period, and are tribal, not Brahmanical, as I interpret the evidence. : The row of human figures at the bottom of the last seal referred: to show a horrilike decoration on the head besides the braid; this might qualify them for the title Vişinin (vii.18.7), while the god of fig. 1 has a headdress which certainly has two (buffalo?) horns for its components. The animals surrounding the deity are to be interpreted as totems, on the great seal of fig. 1.
"
A pol
.... DATA
... .
51
2
Marshall (p:15) misses the significance of the cup-like depressions on the shoulders of the Harappa red stone statuette. They are not meant for fixing ornamental discs, for in that case the little boss in the center would be unnecessary; the intention is clearly to fix an extra pair of arms which could be swivelled around, just as the head is meant to be turned in the neck-socket. Marshall takes the other fragmentary Harappa dancing statuette* as with three heads or faces; though only the stump of a thick neck remains; it had not more than two arms. But the four-armed figure had become so classical as to be given the status of an pictogram in the Indus script. It is rather amusing to' sce Langdon (Marshall,p.446, signs 183,184) leave the particular homo-sign ... ss. coal 53
**** 1944 19
- 1
15.
!!*It might be as well to point out here that the Harappa grey stone image fragment which Marshall takes as an ithyphallic dancing Siva actually represents a young girl dancing. Bronze dancing-girl statuettes have been found in Indus excavations. A comparison of plates LXXX and LXXXI in Vats or the corresponding plates in Marshall will show that the Harappa sculptors could delincate the difference between male and female in every line, not merely in the sexual organs. Also, the seven holes in the neck do not suggest a three-headed image but rather come claborate head-dress or coiffure pegged into place, the head itself being turned to the figure's right. The two holes, below, the waistline correspond precisely to the two bosses of the girdle in such terra-cotta figurines as Vats LXXVII. 51,53; the belted skirt or apron must have been of some different material held in place by pins into the holes.
4

Page #32

D. D. KOSAMBI
unexplained, or call one variety "man supporting two clubs", when an extra pair of arms, or makes, or rivers springing out of the shoulders could be the only possible explanation, as may be confirmed by looking at the corresponding seals in the volume of plates. The reduction to a hieroglyph may indicate that the type originates in or at least is closer to the Indus valley than to Mesopotamia. The transition from the Indus representations of a deity with an extra pair of arms to the Mesopotamian god with rivers flowing out of his shoulders may be seen in Vats, pictograph 383: (pi CXV; and seal 35 of pl. LXXXVI. Possibly, his symbol 358 a might also have developed from the common source. Mackay pl. LXXVI. 8; reports a unique two-faced clay image fragment, the faces being beardless and slant-eyed whence the connec tion (if any) with the two-faced Mesopotamian Usmu is not direct. RV. i.51.5, seadhathir JE adhi śupta: cjuhtala 'those who sacrificed upon the shoulders and were destroyed by Indra' might indicate cults related to the above Indus pictograms or rather to their originals.
52
The absence of fuller archaeological evidence from the Indus valley forces us to consider parallel Mesopotamian seals, permissible because the existence of a common element to the two cultures is admitted*, The Hydra Naga, Sesa) appears with five or seven heads (Frankort p. 72, fig. 26; Pl. XXIII.,j); much later, human figures with two animal heads, goat and stag (ibid.p.271). As the labours of Herakles originate in these seals, the threeheaded Geryon Cacus, or a Kerberos, would have linked up with the Indus seals. However, Ea (originally Enki, a water-god like Nārāyanaj has a twofaced attendant, Usmu according to Furlani, who performs the functions of minister and herald, i.e. is equivalent to a human priest or priest-king. The two rivers flow generally from Ea's, shoulders, occasionally from a jar in his hand. His other attendant, a bearded naked athlete of the Gilgames-Herakles type, also sometimes holds such a river-jar. Frankfort Pl.XXVIII. shows both on a Babylonian seal, in such a way that the rivers might seem to emerge from the hero's shoulders; this seems to be the general case later, c.. PL. XXXIX.i; in XLIV. the river goddesses themselves might be the two attendants flanking the hero from whose shoulders stream the waters. On PL.XLIV.i (a peripheral seal the two streams emerge from a naked goddess's shoulders, as well as those of a much smaller male, perhaps her son. As the water-hero goes back at least to Akkadian times, we must see in him a representative of Ea, and the two-faced attendant must be another such, like the goat-fish which is later Ea himself. This will have to be used in interpreting Indus
*Pather reluctantly, Am Billing to Il Viden 1625), p.11; C.J. Gadd Proc Brit. Acad. zvili, 1232, pp. 121-216, H. Frank Glider Sel London, 1933, pp24-21. My quial thanks are due to be. P.D. Barnett of the Britt. Mevcom. Ese reference, panicularly to L 93115. For the sere antediturian saz, C.L. Worney, JRAS. 1624, pp. 6:3-713; Zimmer, Zeit, für Asyriologie F 25.1924. p. 151. Both Gigam and Batida appear on Ind val

Page #33

ORIGIN OF BRAHMIN GOTRAS
53
vallcy cvidcncc, and Rgvcdic rcfcrcnccs.. Thc goat-fish symbol of Ea (Franksort, PI.XXV.d; XXVIII. k; this sccms to me thic original matsya-avatūra and aja ckapīda ) is rcvcrscd on thc Mohcnjo-Daro 'sacrificc' scal, in that the animal bcforc tlic god is a ram with the head of a fish (first pointed out by Fr. H. Hcras S.J.). The sun-god has, like some other dcities, rays emanating from his shoulders; usually thrcc from cach but thc number is not fixed. This must be the original depiction of saplaraśmi, thc vcdic adjcctivc. Other dcitics
TYR
Fig.pl.
The Threc-faced Indus, God
(After DK. 5175)
Fig. 2. "The Sacrifice"
(After DK, 6847)
MILAN
hannon minus 18
IO
Fig. 3. Resurrection or liberation of the Sun-God from his mountain grave. (After BM 89115)

Page #34

54
.:. D. 'D: KOSAMBI. have vcertation replace the rays (Frankfort Pj.XX.6,4,j,!). On the Gudca scal Frankfort p.143) the dragon-god Kingiszida shovs ivo enakris or dragon, rising from his shoulders, like the later god Tilpak, XXV, 5, nhich relavs both to the Zohak of the Shah Namch. Occasionally, as on the Hammurabi stele, the rays curl up at the end, and 'harassh II:hnaton's rclickstell us that they could: terminate in hands, whence it is natural that they should develrp from or into supernumcrary arms. The best cylinder.scal for our purpose is fig.3, from PM 89115, Frankfort PI.XIX.2, wlúch shows the sun god being resurrected or libcrated from his mountain grave by Ea ani a godde::. !Ishtary. The sun . and the goddess show rays cinanating frota their shouldering inç central r23s of the goddess terminating in what might be taken á hands. Two rivers, proved to be such by the fish swimming therein, strcar out of Ea's shoulders, and he is, followed clo:ely by the two-farci atundánt. As for the godáees, whose various traits are fully; ensugh descopul on cals; bs; the time of the first Babylorúan dynasty to prore her identity with Ishtar, the rays radiating from her and her soling the sun would make her als a daun-goddess. As such,"she has a great deal in Cimmn with the Indian Ligas, wortippos cren: in the plural in the Pgveda, 190 prominent for a more goide of the dawn.' Irixirai comics ints siolent ciflict with hit, heticrirez Hoces car 1.15.6 ; iv. 2016-11; %.138.5., 7.73.6;.tris has, fortunau.ly, lo scal interpretation as a . naiure myth, and can only indicate a clash of cules. If now la were a. mother-goddess for which onccan casily find Bgccic indications; like Ishtar, her bringing out the sun (originally Tammuzi vould suill be remembered after the Aryan conquest and vivuldi cnable her w claim a modicat position as dasunoddess, even after Indra had put bit to flight. It is known that EnkiTo is originally the god of the land, 'st of the waters. Frankfort p.116.fsz. 32 shows us Inanna-Jshter scated as pricatess befoc HCT Ovinirn200-altar, creating homage from some detec; she holds the turriyet jer in her hand. Thus the naked goddes, on Syrian group 1 scals from whice shoulders the two rivers stream is an old survival, and. Ishtar musi-possibly under one other namoha:c been the earlier river.dcity displaced peacefully toy Pa Hot consort Tammuz is beveiled as both huzband and son, the root-7:00 damu meaning both;Tļus is quits natural, and whitever we have a clear historical Curic of develoçment witün the culture, patriarchál cults develop in corriendly tius manner from the matriarchal, by consortship of a son or husband with the priestcas. To revert to the common subytra
cerpts to this common substratum for the Indic and Mesopotamaian river-civilizations, it may be printed out that the horno
be winted-sut that the hoonest headdresurs of esopotamian god, thougti more complicated, again ómnect them with the three-faced Indus' god, as well as the deity' on the sacrifice spal. The
latter scal has seen attendant figures with braided hair, and the numbers ... interesting
ting thoúgh thie;. lack individuality. The seven -sazu7 Isaplarsisi arr not only an Indian group, bar highly reminiscent of this gre at antediluvian sages, whosc images are actually found buricd in group of

Page #35

ORIGIN OF BRAHMIN GOTRAS
ពិត
Marshall (pp.64-5).takcs thc dcity and ministrant figures: in the sacrifice’ scal to bc fcmalc which sccms quite unlikely to me, whilc thc .aninal, is ascribed a 'human'.facc instcad of the quitc: obvious fish, which, argucs, lack of carc in cxamination, or' myopia..... pl.; : .....: : :, ;, ! .:
:.. The row of scien' figures: marching 'singlc filc-hand in. liand, but in the 'oppositc dircction appcars again'on 'à fragmentary scal. (Vats, Excavatious.al Harafipa, New Delhi 1940 plXcl 251): Thic principal difficulty lies in proving thcir connection with the seven Mesopotamian 'ancient apkallı whio wcrc bcforc the flood in Shiuruppaki' Thcir linc of desccht in India is clcár cnough. R:V. ; i.24.10 calls the stars of Ursa Major ykşāḥ, thc Bcars; 'Sat Brāh. ii.1.2.4 mak'cs thc Plciades (kļliikās) wives of thesc Bcar-rşis. Sat-Brüh. vi.1.1.1 cven: claims thati thcsc rşis. wore themselves out with toil.crcating thc univcrsc, which fits thc tkiv. 2.15; Sat-Brāli. ix 2:3.44 tells us that these scvcn wercaddresscd as 'scvcn tongucs, and were made into one person. The idea of our scven primary groups is obviously much older than the beginning of the present. clan system. The scycn; sagcs; as: vipras or, rşis arc callcd: iour ancestors”. Dyıthc Brahmin'secrs of Rgvedic lymns, particularly, by: thc Argirasas: in iv. 42.8 which makes them prcscnt.when.Durgahá's son - (Purukutsa),: was taken prisorcr. and vi: 22.2: taru nah hitaro navaqvā, sapla ,vipräso abhi vājaranta. This might sccm self-contradictory as the Scvcn cannot be split into thc Ninc or the Ton; but association of the scvcn'sages with tlic Navagvas, and. Dasagvas is rcpcated in i.62:4, and perhaps ix. 108.4 whcrc Dadhiañc appcars as a Navagva. At thc very lcast; -wc can say that thcy arc prc-Ariyan associates, ofa.mothergoddess in crcation: The goddess, survives, later as Uşas, daughter of the sky, after being.smashed up by Indrdias an.cvil-plotting: fcmalc: (iv.30,-8,11)... Tlic Mizar-Alcor combination in Ursa Major is still known as Vasiştha and Arundlatí, but we havc scvcral other versions, in which the smaller..companion star is thc coinmon:wifc.or all.scvcn of the 'sages. (Mbh. 1...188.,14): .-. , Pryslys 118.jir. ..4: ! W ... !! Hii 1 1. , 11:,,!!. 1,! :1119. ...,:!' It is clcar, though difficult to prove that 'thclunnamcd scvcn láid low by Indra (x:49:8),and whosc cncmy Indra' becamc.from: liis very birth'though thcy had' till then bocn without an cncmy:(viii. 96.16)arė thesc, seven sages. Thcir súppóscd'consorts, thc Plciades, arc to bc scen ostcn cnough as a constella'tioni on Mesopotamian' cylinder scals. ...;, ., Tus, a, irri siis !121:1., 11:... ..: ill. ; ils, i .:::... . . 1,17 ! ,!
Ugas as a mother goddess connccted with the seven sages appcars explicitly :in iv.2.15 : adhā mālur uşasah sapia viprā jāyemahi prathamā vedhasā nyn ;-'wc:scven sages will gcncratc, men from mother. Uşas who (will become the first ritualists'; we shall becomc Angirasas, sons of heaven, wc shall burst thc rich mountain, shining forth... Motlcr-goddess figurines arc pcrhaps thc .commoncst v Indus city finds, one typc'bcing bird-hcadcd, like the dovc-hcaded Venus of thc carly Mcdítcrrancan culturc. Marshall p.52 describes the scal,on. Plate

Page #36

56
D. D. KOSAMBI
XII, no. 12 which shows a mother-goddess upside down, giving birth to vege
on', hence presumably thc carth-mother, the other side represents her or another femalc dcity scated, with streaming hair, approached by some inalc worshipper. This last is mentioned only becausc Marshall intcrprets the scene as the sacrifice of a female by a man, for which the scal itsclf shows not the slightest evidence. The Angirasas bursting the mountain, a common cnough figure of spcech, is highly suggestive, when we compare the action of the Sungod on thc Sargonid scal, with v.45.1-3. Only the saw is nccdcd to completo the description. But the Rigvedic scenes are remarkably well depicted on Frankfort Pl. XVIII a, where the god of light bursts the mountain and causes the gates to be thrown open.
• One important difference has to be emphasized when considering these resemblances. Weapons such as spear or lance-hcads found at Mohenjodaro have been so flimsy that they could have served only for decorations in some ceremonial ; this contrasts strongly with the sturdy bronze tools found in the same deposits, and with the war-materials in Mesopotamia. Allowing for the painful incompetence of our archacologists, it still seems evident that the mechanism of violence was less developed than one would expect in a city of this sizc, even though it was primarily a trade and manufacturing center.
- The archacological evidence for battle and conquest being undcniable
ne may tenture to identify Harappa with the Hariyūpiyā of vi. 27.5, making thic assumption that the locality has picserved its name through the millennia. The himn praises Indra's shattering the front line of 130 panoplied Vịcivats whercby the rest of the army was broken in the battle on the Yavyāvati river; thuis Indra handed over thc Varaśikhas and Turvaša to Daivavăta, which may bc Srnjava as well as Abhāvartin Cavamāna. Rathore identifications, which can have littic valuc till-we read the Induer attention may again be callcd to the two seals above. The three-horned) trident which the supernal figure wears on his head in the 'sacrifice cea lated to the buffalo-born headdress of the three-faced god in the better
Cloffic. 1. as well as to thc three faces of that god, and the later tria cumbol. The adicctive śrgin does occur occasionally in the Perveda:
seems to bc threc-horned according to v.43.13 and (Agni described as) Trastr secms to bc thrcc-horned according we have noted the Visanin tribe, labcllcd sitāsas, in vii.18.Lastly, anya
astly, anyone with
24 soals of an original mother-golds fron why creatin carne into bring : H i er av for parturition) w for partusium wat
: Sayaga takes this 2n iron which ultrafoda: means 'wiih feet in the ait
lar smal woscof the iniretation on driteful to me, would com t's connect the with the particular sal worrcof the inta the 'vegetation' resernbling 2 craiy.
danta Sakti, persino the : Pafic, Erasmx7.2 s2 that Gauritii, descendant of Sakti, perfstand then
bly the old Rán. This is the only other place
car salira to reach urcat benefits on the banks of the Yayasati (presumably the old Pái, Thirinin
me atyrie. As Darrivata i 2 Bharata, chee 1 harc ten able to find rithont of the two placc-12m" ałynie. Ac Daire
sistent and has a historical appearanc.. Gauricili a Varistha, the wholr: arrunt is ourally consistent and har hírtcsinal

Page #37

ORIGIN OF BRAIIMIN GOTRAS
57
the three-peaked headdress as on the sacrifice scal could be called trisanku, and as the figure is between heaven and carth (probably a god descending for the sacrifice), we have here one posssible source of the Visvamitra-Trisanku myth.
For the first identification of the later cemetery at Harappa as Aryan, cf. V. Gordon Childe, "New Light On The Most Ancient East" (London 1935, 223-4); R.E.M. Wheeler 'Ancient India' no. 3,1947,81 ff, gives a discussion of the archacological evidence for Aryan conquest and occupation at Harappa ; for the ponderous incompetence of Marshall's and Mackay's cxcavation of Mohenjo-Daro ibid.p.144.
IRANIAN PARALLELS
10. There is no doubt that Indo-Aryan society as reorganized with Brahminism opened up the swampy lands of the Gangetic basin, so that caste was an essential feature of more efficient means of production, the development of fixed settlements, and the state. The word brahman for the priesthood is not to be found outside India; and whercas exogamous patriarchal gentes within the tribe or community are known to have existed among Latin and Greek societics after the Aryan invasion of those respective territories, we have no general example of fire-priesthood as the exclusive prerogative of a hereditary caste, though occasionally a gens has the rights of chief priesthood for some particular cult. There is, however, a rudimentary caste system and a fire-priest caste among a neighbouring Aryan people, the Iranians; this case has to be considered in detail.
Our sources of knowledge for the Iranians are the fragmentary Avestan and Pahlavi religious texts, plus the reports of Greek travellers and historians. The first group of documents is lacunary, of late redaction as shown by the reference to the followers of a heretic Gaotema (Yt. xiii. 16, now identified with the Buddha and not Nodhas Gotama), and in addition bears the stamp of à thorough religious reform, that of Zoroaster, which succeeded with the Achacmenids in the 6th century B.C. Comparison with the Rgveda is difficult. Greek notices supply foreign travellers' accounts far superior to anything comparable for that period in India, but are occasionally hostile and sometimes
*I follow: for Avestan sources, James Darmsteter's translation in the Sacred Books of The East, vols.4 and 23 (Oxford 1805). For the general background, Mancckjcc Nusservanji Dhalla's 'History of Zoroastrianism' (New York 1938) seemed to be competent ; for most of the contested points, Herzfeld's discussion in his "Zoroster And His World' (2 vols. Princeton, 1947) seems quite reliable, with a few possible exceptions such as the identification of soma with the vinc. p.551., Herodotos is cited from the familiar translation by Rawlinson, with the abbreviation Her. Other abbrevations: Vd, Vendidad, Yt. = Yast.
8

Page #38

1). 1). KOSAMBI
68
not crcdillo. Taking all thcsc into considcration, thc presence of at Icast onc major strcam of common tradition lctwccn Indic and Avestan Aryans is not to be douhıcd. Apart from thic language of Unc gathấg and old Persjan inscriptons, so similar to Sanskrit, we have the common fear of the demons called pälu, worship of Viyu, love of the sacred haoma = soma drink, and the basic position of the firc cult. Stcady contact had bech maintaincd through regions known to both pcoplc, a3 for cxamplc Vackereta of the cvil sladows' (Kībal), and the land of thc seven rivers (=the Punjal»), the seventh and the fifteenth respectivcly of the sixtccn regions created by Ahura Mazda (Vd.). King Yima is much more prominent in thc Vendidad (Fargard ii) than Yama in thic Rgvcda, but the identity is not in doubt; thic Avestan Sarasvati ("the hcautiful Harahvaiti" of V3.13) is thc Arghand-is, and not one of thc scvcn rivcrs in India. Vcrcthraghna is the "lory made by Ahura Mazda” (Vd. xix.37, and Yt. xiv); Indra has been made into a dcmon by thc rcform, though still under the title of dacva (Vd. xx.13,x.9). Then there is the rathcr ambiguous position of the golde-licclcd Gancarcva, a demon (Y1.v.38,xix.41), but not without respect (Yv.xiii.122,xv.28); lc has been transferred to the dcep though the Indians place him in thc atmosphere. · For our main purposc, we have to note specifically the three supposed castes of the Traninns (Vd.1.16, three races, from the Azerbaijan). But thc division into fire-pricsts, warriors, and husbandmcn is not a degradation of thc Jast as it was for thic Vailya in the Taittiriya Samhita and latcr Indian scrip
r they are descended from tlic tlircc sons or Spitama Zarathustra who is himself not only the first and foremost firc-pricst (Yl.xiii.94) but the first warrior and the first plougher of the ground as well (Yt.xiii.88). The husbandman is honoured on carth, and his progenitor suprcmc in tlıc Var of King Yima (10.j.). Wc Jave thcroforca division into classes, not castes. Now the Avestan title of the fire-priesthood is alhravan, which is undoubtedly thc vcdic atharvan, and again shows an ancient unity of tradition to which Zoroastcr rcvcrtcd in clcaring ost thic bloody (and of coursc uncconomic) sacrifices that ubscurcd thc (supposed) original purity of Aryan worsliip, whcrcaa Buddha and the Tains took up the philosophy of ahimsū. The Iranian Athravan lcads the way alicr a path has been purifica from thic cxtremc pollution of a funcral (Va.viu. 19). Thc Athravans who read, and their pupils, will be and prosperity of Ardvi Srira Anihita (Yl.v. 86). Yt. xiii. 147 says "here are the áthravans of all countrico" come to worship the I'ravalis, whilc Yt.xvi, 17 rcfers to the Athravans sont afar, presumally wandering (cvcn mendicant) missionarics. The finc qualitics of an Athiravan arc given in Yt.xix.53. á the casto still monopolizes the pricsthood among thc Parsis, theoretically endoFamous though not rigidly so in practice.
On tho other hand, wastorn travellers know of Iranian picsts as Magi thoughZoroaster uses magus and magofal only as adjcctives, with the meanin
i the mcaning oi

Page #39

ORIGIN OF BRAHMIN GOTRAS
59
.
grcat. The original Magi were onc of the six tribes of the Mcdians (Her. i. 101), who were a western branch of the same race, first subject to the Assyrians, then independent and overlords of the eastern Persians, and finally conqucrcd by the latter but in closc alliancc ncvertheless after Cyrus and Darius I. Yet the Greek tells us that thc Magi took a pcculiar delight in killing all living things cxccpt dogs and men (Her.i.140) The special protection given to dogs (Vă. xiii ct passim) is, of course, a feature of ancient Persian mcans of production and of thc high status of the husbandman; the dog in the Avesta is the most uscful of man's friends in the protection of the household and of cattle. The killing of all sorts of lower animal life which Herodotos notices is sanctioned, and cvon demanded by Vendidad xiv.5-6. For our thesis, it is of special interest to notc that the Magi rccovcrcd thcir original position of respect, * and continued as an "honorary tribc" to be priests (with readjustment to the new rcforms) but that they had first undergonc attacks similar to thosc suffered by thic Bhrgus and other carly Brahmins. In particular, the story of Darius and the false Smerdis (Hcr.li.61 scq., fully supported by the inscriptions of Darius) and the festival of the Magophonia' (Her.iii.79) show that special action had to be taken against thc Magi as a whole, but that massacre did not
end their pricstly function. For that matter, we also know that some of the , older gods had to be rcadmitted into the panthcon (Herzfeld p.401, 408-9)
though with suitablc changes. In other words, we have a parallel to the happenings in India, and for similar rcasons : conqucst and rcassimilation, with a conquered (though herc Aryan) clan imposing itself upon the pricsthood by virtuc of superior ritual.
The Rgvcdic atharvan, though belonging to so remote a past as to appear morc than human, and without a surviving gotra (unless we infringe upon sacrcd tencts of philology to relatc athar to atri) to commemoratchis cxistcncc, still occupied a far more important historical position than would appear by the comparativcly rare citations. In x.14 and x.21.5, he is associated with Yama whilc in x.120.9 we have BỊhad-diva as an atharvan; but the signal honour given in x.120 to thc supposed rşi and the actual meaning of the name itself scem to rcficct thc stature of somcone likc Ahura Mazda, who is himself a sky-god (Her.i.131) sublimated and an Athravan (Yt.i.12). In x.48.2 Indra Vaikuntha declarcs that he protected Atharvan and Trita, and bestowed upon them the cattlc releascd from ahi, presumably Vợtra referred to as a snakc; an Avestan parallel to the Pani cpisode is perhaps the prayer of cattle to Mithra, for relcasc from the den of the Druj (Yt.x.86). In viii.9.7, as in ix.11.2, atharvan is clcarly the firc itsclf upcn which soma is sprinklcd. In vi.47.24, the Atharvans and Päyu Bharadvāja receive ten special chariots and a hundred
* To the extent of imposing exposure of the dead in spite of original burial (Her.i.140, Herzfeld
147) or crcmation (Herzfeld p.748). Dhalla takes thc Magi as west-Pcrsian pricsts, Athravans as castcrn.

Page #40

60
D. D. KOSALEI
head of cattle from Asvattha; the same Päyu Bharadväja as the supposed seer of x.87.12 mentions the atharan flame as most effecere in driving away
átudhāna demons. The bones of Dadi:;añc tharrara are used by Indra to bill the nine ninetics of his dark cnemies : 1.84.13 indro dodñico asthabhir tstrārijs afratiskutaḥ; jaghāta natalir tata. In fact, Atharvan is cxplicitl; the first gajia sacrificer according to i.83.5 and x.92.10 while the atharsan fire-drill or method of lighting the fire is lauded in vi.15.7 as in vi.16.13-14; the last șk calls Agni by Inára's titles, ustrahanam puran daram, which shows again that fire was used as a poliorcetic vcapon by the carly Irdic Aryans, and incideritally cz.plains how the Avesta could separate Verethraghna from India. The most important of all references to Atharvan is z.14.6 where we have an association in the same line with Anxiras, the pitrs, the Nine secro ratages, and the Phrgus, thc last of whom also appeared in z..92.10. at this stage, we note that the ks containing any reference to Atharsan are overwhelmingly of Bharadvaja or Gotama origin, i.e. of Angiras authorship. Lates, the whole of the Atharva-veda is called the Atharvátairas, (cf.Mbh.5.18.5-8; and the special combination appears with the highest eminence in that reda. Finally, we have seen that the BhrguAngiras combination also crists, which shou's just why thc extinct Atharvan was important in India : The Athartan is the proper fire-priest of one Ar;an group, and association with him was the means whereby the Angirasa; and the Dhrgus climbed into medic triesthood. This gives us much the samc historical de clopment as that of the Wagub in Persia. In the Mundaka Upanigad 1.2 we have the line of teachers as Brahma-Atharvan-argir-Pharadwaja Satyavāha-Argiras. This is a step towards the final inversion to be found in still later traditions which makes Atharvan an Aigiras, the very first.
Hoxe er, not everything can be explained b; parallel histrical developments, and like the name of the riser Sarasvati, there is possibilit; of a legend heing transferred. The story of thc hero Thraetona and the dernon Azi Dahāka is here of considerable interest. The Persian hero of the Athvga clan derforms
reat sacrifice of a hundred stallions, a thousad ozea, and ien thousand lambs Trasna Yt. iz 13-14) or Ardsi Sūra Anānita Yts33-34). Or Vayu (Yt.
0294. Yeryü.33-34 for the destruction of the snake.Az Dahāka himself, "the three-mouthed, the three-ncaded, the six-sjes, v:ho has a thousand senses, that most powerful fiendish Druj, that cemon balcful to the viorld"
e same sacrifice in the land of Bãri = Babylon) to Ardi Sura inahita (Ytr.29-31; and to Vayu (Yt.xv.19-21) "in his accursed palace of inta in order to destroy the severi habitable regrms of the world Karsh
are reiected. The hero Thractona áthwya vares). but his great sacrifices are rejected. The hero Thraa
him and set free his "two vives, Savanghavāc pass successfully to destroy him and set free his "txo vires
who are the firest of body amongst women, and the most
tates in the world" 1.17.14). As a three-headed demon is the Presia as Tribiras Tragira, and in the slaing. Trita inte has
and Erenarāc, who are the fairest of wonderful creatures in the varias vei

Page #41

ORIGIN OF BRAHMIN GOTRAS
been scen to be associated with Indra (x.99.6, x.8.8) which has been taken as sufficicnt for the identification by most scholars. The divinc Vāc, of which thc Rgvcda knows more than onc varicty, though not as thc wife of Triširas, is the specch monopolized by our Brahmins, latcr dcificd as Sarasvati. Thc lcgcnd descrves a littic closer analysis. The Avcstan Thrita is thc first licalcr and founder of medicinc (Vd.xx), but a member of thc Säma family, which again sounds familiarly vedic. Traitana occurs only once in the Rgveda, as thc prc-Aryan or dcmon (dasa) whose blow at Dirghatamas rccoils upon himsclf, Icaving the sage unharmcd, to float down thc river : 1.158.5 siro yad asya trailano vitaksat svayam dāsa uro amsiv api gdha. It is possiblc to scc thc discordant fcaturcs at a glancc; thc grcat difference of territory between thc fourcorncrcd Varona (Tabaristan), for which Thractona was born to smitc Azi Dahāka, and the castern portion of the Indo-Aryan domain is significant. In addition, Azi Dahāka survives to tcmpt Zarathustra : "Rcnounce the good Rcligion of thc worshippers of Mazda, and thou shalt gain such a boon as Vadhaghna gained, the ruler of the nations" (Vd.xix.6). Yct the historic substancc of thc lcgcnd is cnhanced by analysis. In the first placc, Azi is a king, as shown by his palace and grcat sacrifice, which was not only repeated by his slayer but (ncar lakc Urumiah= Caccasta) by Kavi Husravah, "He who united the Aryan nations into one kingdom" (Yt.v.49,32, ix.22); at the Whitc Forcst by the 'murderer' Aurvasira ficcing from Husravah (Yt.xv.31); and by Xerxes (Her.vii.43, 113; cf.i.50). His connection with Babylon is curiously supportcd by latcr lcgcnd, for the Shah Namch describes him (=Zohäk) as with two snakes springing from his shoulders (cf.p.27 of thc Shah Nameh translation into English versc by A. Rogers, London 1907). Zohāk is not an ordinary king but a successor to Yima-Jamshed himsclf. The black snakes that issucd from his shoulders (as the devil kissed him therc) appcar on Mesopotamian scals as shoulder-rays from the sun, dragons from the shoulders of Tispak-Ningiszida or rivers issuing from thc shoulders of Ea or the hero Gilgamcś of Sumcrian lcgcnd; from them to the four-armed characters of the Indus valley scals and later sculpture is only a step, the actual transition probably being in the oppositc direction. Siśupāla (Mbh.2.40.1) the Cedian was born four-armed and thrcc-cycd. The god (?Sun) on Hammurabi's stelc has curved slamcs issuing from his shoulders. Thus, the legend is rooted dccp in thc historic tradition of Aryan conflicts with grcat prc-Aryan civilizations in thc Indus vallcy as well as in Mesopotamia ; we know that thesc civilizations had long, continuous co-existence and intercourse, as well as many commom fcatures, probably some common origins. In this case, I should be inclined lo consider tlic event itself as having occurred in thc Indus valley.
Just what the snakc-dcmon significs I cannot venture to say in this context, cspecially as his connection with the cult of the Mother-Goddess and pro-patriarchal family life is known, but not prescrvcd in cither of the two Aryan sources under discussion. However, other hcrocs conquer multiple-headed

Page #42

02
D. 1). KOSAMBI
snakes as for example Herakles and the Hydra, or the Indian counterpart Krana and Kaliya; yet the Hydra has one head which is immortal, and Kṛṣṇa only subducs Kaliya without killing him. The vast though inobtrusive current spread of the naga cult need not be given in detail. One major Hindu holiday is dedicated to the cobra. Cobras are regarded by many (my mother, grandfather, uncle, and cousins among them) as embodying ancestral spirits, and the live snake himself generally forms an appendage of most rustic temples. Sesa's bearing the whole earth on his multiple hood goes back much further than the obviously recast legend in Mbh.i.32. Visnu sleeps upon the great (many-headed) cobra, Siva wears him as a necklace, and the cobra's protective hood is reared above the phallic symbol of Mahideva. The chief cobra Taksaka escapes being burnt down with the Khandava forest (Mbh.1.218.4; the whole episode is one of land-clearing in the typical Aryan manner, by firing the woods and slaughtering all inhabitants), and is barely saved from Janamejaya Päriksita's firc-sacrifice by his human newphew Astika. The name takşaka is related to takşan carpenter, hence to the god Tvasty; and to Taksasila, (the Greck Taxila) which was the capital after the Mahabharata war. Thus Taxila to Kurukṇctra must have been the territory of a tribe or tribes which had a cobra totem or cult. Nägas remain extraordinary craftsmen in Indian folklore, demonic beings able to assume human form at will. Krana's elder brother is usually taken to be an incarnation of the great Naga. The demon Vṛtra is called ahi in the veda, but the snake of the deep ahir budhnyas remains an object of worship. References to ahi are scattered throughout the veda with the important exception of the Visvamitra book. Here, the word ahi is found only twice (iii.32.11 and iii.33.7), in both cases referring to Indra's killing of the demon Vṛtra in order to release the waters. The peculiar difference between Vrtra or Trisiras and Azi is undoubtedly to be explained by the historical differences in the relations between the Aryans and the conquered people in India, as against the Aryans and their Assyrian enemics in Persia. As for the Angiras Dirghatamas (Bṛhaddevata iv.11-12), his name itself shows association with darkness (explained away by his blindness, i.147.3; iv.4.13), hence with the Vṛtras who are the enemics of Indra and the Aryans. But in spite of the familiar royal persecution he left descendants who became Brahmins in the main priestly lincage, while Traitana left his mark only upon a very distant branch of Aryans. Thus even this legend supports the contention that the development of Indo-Aryan sacerdotal tradition is by assimilation of a pre-Aryan element, which has special connection with the Brahmin caste, particularly in its original stages. With the Zoroastrians, success meant that the religion was predominantly that of a comparatively small number ruling over vast territories inhabited by far more numerous peoples which had diverse customs of their own and in some cases law-codes going back to Hammurabi. Therefore, the development of a new gotra system among the Magi was not necessary. In India, on the contrary, the conquest meant destruction of the

Page #43

ORIGIN OF BRAHMIN GOTRAS
03
Indus vallcy urban cultures, rcorganization of socicty into castes, and progressive opening up of now, sparsely settled, and hcavily woodcd tcrritorics to the cast. This gave opportunity for cach group of pricsts to be attached to or aclopted by scvcral Aryan clans, which must have been thic origin of Brahmin pravaras.
TRITA ĀRTYA; THE ORIGINS OF EPIC AND SAGA
11. The Avcstan Vadhaghna can be cquated without difficulty to Indra himself under thc title of vadhasnu, bcarcr of the death-dcaling weapon, though vajrin, vajra-hasta, ctc. arc thc usual adjcctivs. Vadhásnu is actually uscd of Soma (=inlu) in ix.54.3. Wc hicar of the gods shooting down upon men (v.41.13) and Agni brcaking down walls (vii.5.5) with a vadhasna. Indra's weapons have thc same name in i.165.6 (vadhasnaiḥ); cquivalent forms vadha, vadhar, etc. arc found in considerablc profusion : vi.83.4, Indra-Varuņa vadhanibhir vanvanti ; So, Trikiras bcing a purohita of Indra might be rcficcted in the association of Azi Dahāka with Vadhaghna in thc Avcsta.
or the block of seven hynus (1.51-57) ascribcd to Savya Angiras and all dcdicated to Indra, 1.53 begins "Let us sing a hymn to grcat Indra, dcdicatc chants to him in the abode of Vaivasvata”. The location is reminiscent of the Var of Yima. "Thou (Indra) hast crushed under thy irresistiblc chariotwhcel thc twicc ten tribal kings with thcir 60,099 mcn, who fought against kinless (abandhuni) Subravas. Thou didst aid Susravas with tly support, Indra ; with thy protection thou gavest to the victoriously advancing (turvayīnam*), Kutsa, Alithigva, Āyu into the hands of the grcat young king" (1.53.9-10). I suggest that this fits thc Avcstan Husravah very well, though here the v titlc of Kavi is not mentioncd, and thc opponcnt Aurvasāra is not recognizable.
Even morc instructive is the serics of references to Trita Aptya. Lct us first report what thc meticulous Grassmann (col. 557) has to say : "Trita is originally 'the third and therefore set up against a 'second' (viii.47.16).1) Dcsignation of a god who is probably obliged for his name and worship (1.187.1 ; i.163.2.-3; 1.52.5 ; viii.7.24) to a pic-vcdic point of vicw, bccausc of which he also occurs often in thc Zend. Alrcady in thc Rgveda, his original being appcars obscured, in that hc shows to a certain cxtent as the background for thc world of vcdic gods. Thus he appcars in å definitc manner as the predc
* I trcat I trvayāna as an adjective, without yielding to the temptation to take that and the allicd luralyn as “T'uranian". It is an adjective of Agni in i.174.3, of Cyavāna in x.01.2. It seems to be a namc by itselfin vi.18.13; that k rcpcats the substance ofi.63.11 abovc. without the name of Susrayas. Sayana turns tlic meaning complctcly around and makes Indra hcrc thc protector of Kutsa, Ayu. Atithigva. Onc may comparc X.10.3-0, 8 whcrc the same characters and a Savyn) ppcar wiilc 3.8'spcaks of Indra hclping Atithugva against Karañja and Parnaya. Velankar, in the Ann..Bhan. darkar. O. R. Inst. xxiii. 1842.057-008 Con Divodäsa and the other Atithigvar) identifics Kutsa with Ayu and Atitbigva for thc hymn.under discussion, whilc maling out a good cayc for more than onc Atithi ya and several Kutsas (which latter is clear, thic name being representative of n tribc).

Page #44

64
D. D. KOSAMBI
cessor of Indra, who strikes down dcmons just likc lum and frees the imprisoned strcams; for this rciationship i.52.5 is particularly characteristic, where it is said of Indra that he brokc thc dcfcnces of Vala like Trita. 2) So hc blow's upon Agni (v.9.5 ;x.46.3), discoyers him, establishes him in thc houses of men. 3) Hc leads Varuna-Soma to the sca (ix.95.4) and cven sccms himself to be Varuna (viii.41.6) 4) Hc appears in alliance with other gods (ii.31.6 ; ¡1.34.10,14 ; v.54.2 ; viii.12.16), namcly also 5) with the winds (x.64.3 ; X. 114.4) and 6) with Soma (ix.32.2; ix.34.4 ; ix.37.4 ; ix.38.2; ix.86.20; ix. 102.2, 3 ; ii.11.20), so that thc fingers that purify thc Soma appcar as Trita's virgins (ix.32.2 ;ix.38.2), thc Soma stonc as Trita's stone (ix.102.2) and Soma as coming to Trita (ix.34.4). So he is represented 7) as living in thc far unknown distancc (1.105.9) and thereforc 8) carried away to Trita (viii.47.13, 17) is equal to carried very far away. In all thesc conccptions, he appears with the qualification äplya, as also in mcaning 9. But hcsidcs this conception of Trita as a higher dcity, he appcars also 9) as a lower god (1.102.1; 11.11.19; x.48.2 ;x.99.6 ;x.8.8) who performs labours in the scrvicc of Indra or 10) calls upon the gods for help (1.105.17 ;x.8.7) when fallen into a well. Finally 11) in the plural, a whole class of gods is so denoted (vi.44.23) in whose abode Indra found the nectar of immortality".
This shows that Trita, though faded, had at one time a substantial following. The whole ncxus can very well hc cxplaincd by our present hypothesis if the course of historical devclopment hc taken into consideration. Onc may remark that viii.47.13-17, wherc evil demons and nightmares arc exorcised away to Trita Āptya need not just mcan driving them away to a far distance but may also be in the nature of a curse upon Trita. In any case, Trita's distance in time and place from the rgvcdic sccrs and the major strcam ol tradition need not be doubted, particularly as hc finds no mention in the Visvamitra. Vāmadeva, and Vasistha books. The higher forms of Trita must iridicate his antiquity and ancestral position for some clans, say thc Aptyas, whilc thc prayer from a well might prescryc a memory of his actual humanity. Very significantly, Indra is himself called äplyam äplyānām (x.120.6). Knowing what we now do of the Aryan invasion, it sccms plausiblc that Trita is Intra or onc of thc invading Aryan chicfs, latcr collectively dcifice under the title of Indra". His scparation from Indra is helpful, sccing that some time after the conqucst Indra has to be worshipped by brâhamaņas in spite of the stillremembered killing of their ancestors, and destruction of their gods and cities. In fact, we have seen from thc Avestan tradition that Azi Dahatic liscroll
as the husband of two kinds of vêc ; thc word vacaspali is used without
vacaspati as thc husband of two
e don A.A. MacdningH, JRAS.xxv. 1892 pp. 119-196, identiAvaigut my intcrprctation of Trila, scc A.A. Macdninstell,JRAS.xxv. 1802 This with Auni in the same vein, 1. Fowler JAOS vol. 67-1947. p. 39- 130
at Trita is a dubic of Indra alleast in the one performance that infinite
.. XXV. 1892
no posible doubt that Trita is a douhicofin Oi
!. Sams vein, 11. Fowler LAOS . namely the Killing of Tribiras,

Page #45

ORIGIN OF BRAJININ' GOTRAS
០៨
further definition in is.26.4,ix:101.5,x.166.3, whilc wc 'lavc vīcassialim visvakarmiīnam in x.81.7. Vacaspati is peculiarly BỊhaspati or Bralımanaspati, and so it is not surprising to find Brhaspati as willi scven (instcad of Trisiras's thrcc) mouths, sapleisgrasas in iv.50.4, whilc iv.51.4 has jeni navague argire daśague sajlisye rcvalī - revad uşa. Bralima naspati may have developed later (cf.x.68, Brhaspati rivals Indra's scals ; Brahmanaspati as thc crcator, x.72.2) quite naturally into thc four-licaded Bralimī, which confronts us again with thc possibility of purely internal growth: But thcarchacological cvidcncc pointed Lo above, and what is known of thcogony in gcncral, would make it ćštrcicly unlikely that a multiplc-licaded god. was invented out of nothing by the Bralimin class as their own special crcator.' Thc alternative intcrprctation is that onc aboriginal Brahmin god at least survived in thcir mcmory, and was rc-adópted into the ricw panthcon after the pricsts had bccomc“ Aryanized. The Brahmin dcmon Rāyaṇa killed by Rāma had sproulcd as many as ten lịcads !
: Brhaspati is not the only god to grow out of comparatively brics mention in thc Rgveda into quitc overpowering glory: Vişņu is a known cxamplc, and, Puruşa in x.90, cvcn morc striking as Nārāyana. Thesc arc clcarly forcign. additions to Aryan cults, but a parallel to Břhaspati is better sccn in Prajapati. Hc begins, as an adjcctivc, bcing Savits in iv.53.2, and Soma pavamina in ix. 5.9. A cow, has been given by Prajapati in x.169.4 and x.184.4 addresses to him part of a prayer for oflspring. The very late x.85.43. shows him as a god. An cntirc hymn is dedicated to him only in x.21, whcrc hic is mentioned in ilic Jast rk by namc; latcr comment has made the interrogativc kaḥ of thc rcfrain into a namc for Prajapati, perhaps from ancicnt memories of the significance of thc word as a man's soul or csscncc (as it also was in Egyptian). The crowding into the last books is clcar proof of a latcr.date than for Bșhaspati.
Memories of Brahmanical adoption of strange .ways in distress survived quitc latc. We know that the ascctic tradition in India gocs back to period far carlicr than that of the Buddha, and that many of thcsc ascctics were specially lcarncd, as well as verscd in the mystcrics. For a development purcly within thc junglc, this would be impossible. On the other hand, if so'mc of the (originally) unassimilated and uncnslaved pricstly survivors of the prc-Aryan culturc took to thc sorcst and cked out a painful cxistcncc on the margin of slowly growing scttlements, the high rcspcct accorded to ascetics is explained, as well as the gradual merger of the two strcams in latcr philosophy. Manusmrti 10.108 spcaks of Visvāmitra acccpting dog mcat from the hands of a Cāņdāla, but there is no vcdic support for this, and as thc book is of Blirgu redaction (Ms.11.59-60), wc may pass this by. The two previous slokas arc confirmed. Ms.10.107 proclaims that hungry Bharadvāja, with his son, reccivcd many heads of cattlc from Vidhu Takşan. Thic rcfcrcncc is found 'in thc Rgvedá (vi.45.31-33) in a genuinc Bharadvāja dānastüti of king Brbu, i

Page #46

66
D. D, KOSAMBI
the most gcncrous of princcs, who victoriously achicvcd chicftainship of the Panis like Urukakşa Gūngya. . When wc rccall that the Rgvcdic Panis arc regularly maligncd as grccdy, mcrcantilc, and cvcn cattlc-stcalcrs (x.108 gloss) Vstra himself bcing a Pani at times, or that they arc dcmons-which mcans old encmics of the Aryans, it is clear how Bharadvaja had sinncd. However, hc had another Angiras prcdcccssor, Vāmadeva. Ms.10.106 tells us that starving Vāmadeva was unstaincd by cating dog's flesh, and this is again supported by a tk of Vámadeva (iv.18.13) where the sccr narrates that in distress he was reduced to sccing his wisc in degradation, and to cooking a dog's entrails. But this is no less à dānasluti of Indra, in older form, than Bharadvāja's to Bțbu; for Vūmadeva concludes that then thc Falcon (Indra) brought him Soma. This can only mcan granting the right to Soma, which implics granting membership in the trihc, i.e. adoption as an Aryan follower or pricst of Indra. Says Vāmadeva (iv.24.10) "Who will huy from me, for ten cows, this Indra who is minc? After hc has defcated thc cncmics, let him return (Indra) to me". This has been interpreted as hawking an image or fetish of Indra for hirc, a unique practice in thc Rgvcda not supported by any authority. But hiring out the rşi's scrvices for a specific occasion, to secure the aid of Indra 'in battle, would sccm far more natural, would fit thc context of the hymn better, and is also thc traditional Brahmin practice. Getting Indra back is csscntial ; "What use to you (Indra) arc the cows of the Kikatas” (iii.53.14) shows such an attempt at cnticing Indra away from others. As for the specific mention of those who did not bclicvc in Indra, we have two quite distinct classcs : those who arc thc cncmics belonging to thc aboriginal population (vstras, dasyus, etc.) and thosc who are treated with more circumspection though denying Indra, as in ii.12.5, viii. 100.3. Thesc might be vrátyas, cxtravedic though Aryan, but later tradition like that of the Brhaddcvată says cxplicitly that the reference is to particular scors, Brahmins who had once dcnicd Indra and then "scen", i.c. acknowledged him. There is no rcason to doubt 'this, and it supports our main contention.
This talc of woc, being found in all layers of the Rgveda, is no later invention ; x.33 begins as a song of hunger by one who has barcly cscaped dcath by starvation. The numerous dánastulis cannot be scparated (as donc so often by Grassmann) from thc hymn proper. In thc first place, similar praisc is found in thíc body of other hymns, in the same mctrc. Secondly, Malinowski's experiencc with Trobriand Islanders' folklorc shows that the coda is an integral part of the story, primc cause of its prcscrvation. The record of gifts to the sinster could have hccn important only if thcy wcrc comparatively rarc, lifcsaving cvcnts whosc chanting was at once grateful remembrance and incentive to other donors. The properly historical names of the Rgveda occur for the greater part in such dänastulis. Onc can scc groups like the Bharadvájas and the Kanvas cast about for protectors among all sorts of chicstains. Even

Page #47

ORIGIN OF BRAHMIN GOTRAS
67
the two dāsa chicks Balbūtha and Taruksa are praised to the utmost by Vasa Ašvya, and it is their gencrosity to him that, presumably, brings them under the i grace of Indra and Vāyu in viii.46.32. This, incidentally, shows that Brahmin-'. ism cannot be a purcly Aryan growth. Thus the hostility to Yadu-Turvasa, (vii.19.8) and friendship in vi:20.12 are explained because Vasiştha and Bharad-1 vāja were then priests to different, hostile tribes, and called upon Indra to support thcir own party. The all-importance of giving to Brahmins, so nau
scatingly familiar to any rcader of classical Sanskrit, goes to iv.50.7-11 (which .. would fit into any Purāņa.) and is the economico-thcological basis for the pricst's special sanctity and development into a caste apart.
Most important of all, thcsc appended verscs of gratitude provide the transition between fixed, sacred hymn, and improvised, fluid', popular lay; hence the delibcratc change of mctrc in the danasluti. Thc Mahābhārata epic, for example, is a rc-cdited collcction of such lays about the main theme of a great civil war. Every digression (particularly gencalogical) called for by any of the characters is made at oncc, which is clear proof of improvisation. The prologue has a vedic hymn to thc Asvins (Mbh.1.3.60-70; not out of place in the context) and claims that the work is a veda, which could hardly be admitted on the strength of a solitary hymn. Onc may therefore conclude that the glorifications (māhātmya) which intersperse lhc various cpisodes, telling of immcnsc merit to be gained by listening to the particular story recited, make up for the disappcarance of other hymns with which the minstrel must, in older days, have begun his sct portion ; thc māhātmya is a later guarantec that the sanctity originally provided by the hymn has somehow been preserved. The Mbh being of Bhrguid rcccnsion, with a fragment surviving of a rival compilation by Jaimini, we have hcrc another cncroachment by Brahmins; the professional bard: (sūta ; onc actually recites the extant Mbh. according to the work itself), is of mixed castc-son of a vaiśya by a kşatriya woman--which points to an ancient rcspcctablc origin. of the guild, before class differences had developed into impassable caste barriers. Thc idca of caste-mixture is the Manusmệti method of enrolling such guilds into the caste system. The cheerful poct of ix.112.3 says : kārur aham tato bhişag upala-prakṣiṇī nană, 'I am a hymn-composer, father is a herb-doctor, mother grinds corn', all as professionals, for profit ; this is certainly not thc Manusmţti idca of a family. The irrcgularities of Mbh trisțubh mctic approach the vedic rather than later classical models. I suggest that the long tradition of free improvisation accounts in greater part for the 'fluidity of the cpic text as compared with the rigidly fixed veda or Pāṇinian aș țādhyāyī, though all thrce were orally transmitted for a while, and the two last for å much longer period than the growing epic. Vyāsa's stepping out of the role of poct to direct the actual characters of the epic may indicate some sort of stage-direction and the acting of scenes to accompany the recitation ; this would account for the miming of Bhārata-yuddha episodes in Balinese tradition, dérived from South-east India.

Page #48

D.; D.: KOSAMBI ;
TRIŚIRAS AS PURE : MYTH. 51. 12:. Tlie proposition must now be considered that all Rgvedic stories are purc'myth, from which no histórical information is to be derived. The very survival of a myth indicates thc cxistence of a class of people interested in repcating it till such time as it camc to be recorded. Gcncrally, in primitive societies, this implics connection witli ritual and thc priesthood that survivės by performing that'ritual. Thc cxistciicc of an carly written version of the Rgveda is extremely unlįkcly, though not absolutely impossibic , Indic as well as extra-Indian Aryans had had violent contact with ancient litcratc 'civilizations. Writing was unnecessary at the intcrmcdiate pastoral and pioneering stage, from which settlements gradually arosc to dcvclop into kingdoms of an entirely different typc. The priesthood was all the more necessary, and there is no reason to doubt thic gencrally accepted theory of an entirely mncmonic transmission of the oldest veda in its carly days. . The point, however, is not material in our casc.
Identification of ancient city ruins in the Indus vallcy with Dasyu cities ''destroyed by Aryans can no longer be stigmatized as Euhcmcristic. Thus,
the ritual that developed at the earliest period could not be the YajurvedicBrāhmaṇic rite but something connected with, or insucnced by, these clashes. The later vcda preserves little or no tracc of this, cvcn in symbolic form, simply becausc thc-social, political, and cconomic situation had changed.completely, Looking specifically at the story of Indra (or Trita-Thraetona) striking off the three hcads of Tvāşțra, we find its narration and survival almost a casual feature of the Rgveda. Latcr vcdas use it only to introduce more prominent stories, such as the killing of Vitra, which motivate purification and Soma ceremonies. Therefore, the initial ritual, if any, has faded. Yet we have the thrce:or fourfaced god and several three-headed bcasts on Mohenjo Daro scals, as well as broken images with a human torso and one or more head-sockets. Moreover, the trimūrti continues to this day, with a totally different thcology, as representing a deity synthesized from threc later gods, of whom the four-hcaded Brahma is. onc(though allotted ynly one of the three hcads). Finally, there is now no · striking off, the heads of the image, which shows that both ritual and myth;
follow changes in the relations of production. If thc. Tvāştra story indicates, on any. Aryan ceremonial, it can only be the killing of a priest by the king, for
priestly gentes continue to derive their name from Tvāştra, even from his severed heads; the line of descent from Brahma at the end of Bșhadāranyaka Upanişad: ivshows two Tvāştras. But the only other such parallel story is the, striking off a horse's licad from DadhyañcĀtharvan:(also in that line of descent), which head continues to be immortal and prophetic in lake Saryaņāvant, and from which perhaps.Indra fashioned a powerful weapon, like Samson from the, jawbone of an ass (Brhaddevatā iï.22-23; Rv. i.84.13-14 ; Sat.Brāh.xiv. i.i.18-25). This is the exact opposite of what has been propounded about such

Page #49

ORIGIN OF BRAHMIN GOTRAS
69
myths! .: That they represent thc pcriodic sacrifice of a king. Hcrc, instead, of the pricst sacrificing the king, it is thc god-king who bchcads his own pricst. This cannot bc taken as yet another Brahmin inversion, for the vcdic pricsthood grow stcadily in power, and there is no rcason for it to have taken a stcp against its own inviolability. The killing of Vặtra might conceivably be related to a! pcriodic human sacrificc, sccing that urlra also denotes dark non-Aryan cncmics: whence somc ritual for victory over them, or sacrifice of prisoners after a battle, would not be unlikely. For Tvāştra, no such explanation scems to be possiblc.
Study of thc Iranian counterpart Azi Dahaka shows us that we have to do with a non-Aryari king or pricst-king. Thc motif of an initially monstrous king is strong cnough to rcappcar in India down to Sisupäla, king of a historical pcoplc Ccdi. He is thrcc-cycd, which is rcally cqual to thrcc-licadedness, as will bc sccn, and four-armed at birth ; killed by a later god, the dark Kršņa, after many trespasses have been forgiven. It is possible to concludc, following the rcasoning of those who favour such analysis, that the myth portrays, in its initial · slages, the killing of a pre-Aryan priest-king somchow connected with the later Indo-Aryan pricsthood. The killer docs not succeed to, but relains, sovereignty over thc Aryan panthicon. There is nothing like a sacrcd marriage connected with the story, and the patriarchal sociсty of the Rgveda docs not allow anything of the şort to be fitted in. Later antagonism between kşalriya and brāhmaṇa can cxplain ncithcr thc forination of the story nor its Iranian version, supplying at most a causc for its rcpctition, or for the usurpation by Brhaspati of some of Indra's saga. Thus, the "ritual” is at best adopted from thc prc-Aryans, which would normally imply adopting some of thic priesthood therewith.
It sccms much morc rcasonable to admit what has already been demonstrated for Grcccc: That conflict between gods indicates conflict bctwccn two
1A.M. Hocart: Kingship (London 1927); Lord Raglan : The Hero (London 1930). I am sorry to say that Hocart's cvidence comes from a much later (for India) period, and has been reported in a fashion that nccds correction. Raglan's analysis also sccms incompletc, for I can show from personal cxpcricncc how rcal historical characters have had anyths attached to their names without any corresponding ritual or drama to account for the transfercnro of older stories. Attention has to be paid to the class of pcoplc among whom the myth is current, and also to the pre-cxistence of a written tradition, or of other classes, which may provide the raw material for folklorc. Yet thcsc two works contain much that is suggestive and valuable, in contrast to the works of dillusionists like W.J. Perry,
2 Gcorgc Thomson : Aeschylus and Athens (London, 1944); Studies in Ancient Greek Socicly: The Prehistoric degean, (London 10:10). But the direct analoguc is not possible with thc matcrial we are now discussing. Indrāni, the wife of Indira; is a very late addition to the Rgveda, and the great femalc dcities like Durga Parvati, Laksmi, ctc are much later. Umā in the Rgveda docs not appcar to have any connection with the later goddess whose physical merging into the hermaphrodite Siva indicates just what was shown for Grcccc, sccing the positinn she still occupics as Durgā, an castern mother-goddess. The femalc dcitics of the Rgvccla appcar negligiblc, or local, like the dawn goddess sā, the goddess of birth Sinīvāls, or the river godclcsecs lcd by Sarasvati. I suggest that at the carly stage, thc invaders had an overwhelming victory. Only latcr did they find it necessary to admit thcsc older clcments, along with the people who prcscrved that culture or its remnants. Otherwise, we should have a course of development the very reverse of that gcncrally found, from the patriarchal back, to matriarchy. Then, why thc lcast Aryanized of India's primitive tribes have the matriarchal system would be difficult to explain. . My suggestion would also account for the fact that many very old legends, such as those connected with the flood, appcar only at the pogl-vcdic stage. What synthesis lics back of the multiple-licaded Indus vallcy images cannot be analyzed from available sources, but undoub. tedly, they had composite dcities also. My own explanation follows in the next section..

Page #50

770
D. D. KOSAMBI
w
or morc cultures which were thercafter fused. In India, this fusion did not go to the extent of continuing the urban lifc of thc older period, though that was cssentially what other Aryans did further to the west. Had the amalgamation achicved nothing more than the formation of a hclotage (thc sūdra caste) from the conqucrcd black Dasyus, there would be no such indelible mark Icft upon the Brahmin priesthood and tradition. Morcover, there is ample cvidence for the existence of dark-skinncd Brahmins in antiquity, the possiblity being also admitted by Buddhists (Digha Nikāya 4) but not by Brahmins from the northwest (7BBRAS vol.23, 1947, pp.39-46); such clcar cvidence of racial admixture did not lead to any loss of caste. This completes the alternative line of reasoning, bringing us to the samc point as before.
SURVIVALS OF MOTHER-RIGHT IN THE RGVEDA
13. The question of matriarchy* and group-marriagc has only becn skirtcd in the previous sections. I now propose to show that even in our oldest available documents there exists clear cvidence to support our arguments, without violence to logic and with improved meaning. Such re-interpretation is necessary as the original simple meaning had become incomprchensible in the intervening millennia of a totally different form of society. Following the vedas, cpics, purīņas, gļhya-sūtras, and smstis in chronological order, we find at times a reversal in the accepted scqucncc of development. Matriarchal features appcar later, as for example strīdhana (property inherited in the female line), and recognition of consanguinity on the mother's side. These are duc not to retrogression in the means of production but to absorption of the remaining pre-Aryans by comparatively peaceful methods. Matriarchy and the most primitive forms of exogamy are known to survive only among thc lcast Aryanized of India's tribes. The leading Rgvedic gods Agni, Vāyu, Varuna, Mitra have no real consorts, for Varuņāni, Agnāyi etc. (like the male Sarasvat for Sarasvati) are palpablc fictions which ncvcr took hold; the noticcable fact is that they should have been thought ncccssary at all. The slightly better drawn Indrāni (x.86) never establishes herself in the panthcon. Vişnu develops his supreme importance only in the latcr pcriod when he has already marricd the
ahorn Laksmi. Siva-Rudra can becomc the great god because of his wife Parvati: he has often to appear as a hermaphrodite assuming half her body,
Polis her cult. The conclusion is irresistiblc that these divine marriages only represent the fusion of the invaders with a sct of predominantly ma
This is treated to some cxtcnt (1or modern Dravidian India) by O.R. Ehrenf
an India) by O.R. Ehrenfels : Mother-Right
. M. 1941). In India (Oxford 1041). The autior's cit
The autior's citations of our oldest sources are perfunctory, second-hand,
ause of conscqucnt misinterpretation. od irrelevant or inaccurate because of conscqucnt misinterpretation. The
The comparison on pp.180-81 What Marshall imagined to be the essential features of the Indus Valley cuine
have been Nayar civilization at is height is particularly superficial and misleading. Ehrenfels belicvcs to have been Nayar civiliza the supposed features not being exclusive.

Page #51

ORIGIN OF BRAHMIN GOTRAS
triarchal prc-Aryan pcoplcs, but cven that thc absence of such cult-fusion helps Buddhism push the older unmatcd vcdic gods into thc background, in spitc of the grip maintained by vcdic ritual. For direct rcfcrcncc to an carlier stage without forbidden dcgrccs of marriage, wc sccm to have Ait. Brīh.iii.33.1 which spcaks of cvcrything as crcated out of thc incest of Prajapati with his own daughter. The incest, without naming Prajapati, gocs back to RV.x. 61.5-7, and must be much older. Ait. Brāh.vii.13 cvcn says, "thcrcforc a son his mother and sister mountcth," though such promiscuity must have belonged to a distant and repugnant past of the contemporary Aryans as shown by thc Yama-Yami dialoguc. Thc sun-god Pīşan is called 'lover of his own sister' in vi.55.4-5., whilc the gods actually marry him off to the sister Sürya · in vi.58.4. Both thc Achacmcnians and thic Sākyas had traditions of brother-, sister marriages. In thc Rgvcda thc minor caninc goddess Saramā (x.108; 1.62.3; 1.72.8; iii.31.6; iv.16.8; v.45.7-8) finds stolcn' cows as messenger of Indra. Thc tcrmination mā was not understood by the later pricsthood cxccpt as a ncgativc injunction, depriving thc namc of all mcaning. But thc list of female. dcitics or dcmons whose names so terminato incrcascs immcdiately after the Rgvcdic pcriod : Uma, Rusamā (Pañc. Brāh. xxv.13.4), Rumā, Pulomā, Ramā, Halima (Mbh.3.217.9) cto; they arc undoubtedly mother-goddesscs2 at onc stage of thcir mythological cxistcncc. In x. 40 thc lcviratc is clcarly mentioned: ko vá Sayulrü vidhaveva devaram maryam na yoșā krņute sadhastha ü, but the very word ? for widow and the institution of widowhood shows us that thc Aryans had long shaken off their own traditions of group-marriage and mother-right. Thcrcforc, the direct references from the Rigveda which arc cited in thc following paragraphs arc much morc likely to represent absorption of prc-Aryan custom than an uncallcd-for reversion 10 ancient practicc.
My main argument is the following. A single child with many mothers is characteristic of a society in which group-marriage is the rulc. "A child gives the namc of mother not only to her who borc him but also to all his maternal aunts. A Europcan not familiar with thcsc relationships is surpriscd when he hcars a native (of New Britain) boasting of having thrcc mothers. His confusion is incrcascd when the alleged thrcc mothers stoutly asscrt ‘amilal ga kava iva, all thrcc of us borc him”. This is quoted from J.G. I'razcr's Tolemism And Exogamy, (London 1910, vol, 1, p.305, focinotc), bcing itself apparently taken from P.A.Klcintitschen's Die Küstenbewohner der Gazelle-halbinsel. Wc shall now proceed to show just this attitude in somc hymns of thc Rgvcda.
* Sarama's tracking down cattic stolen by thc Panis is unqucationably a later story, to explain a lengendary strisc. No Rgvccic hymn which refers to Sarama says anything about the cattle having lccn stolcn. The goddess presents a blunt, aggressive demand from Indra to the l'aņis, apparently for their own cows, in x. 308. The other rcfcrcnices gcncrally show that 'cows' can bc understood as rivcrs; best of all in vi, 10.8.
* For Mä as a mother-goddess, cf. Amarakośa 1.1.29 ; what conncction cxists with thc Hittite goddess of the same name is not known.

Page #52

. D. D. KOSAMBI
It may be objccted that a plurality of mothers may indicatc only polygamy, A moment's thought will make it clcar that in a polygamous gentile patriarchal sociсty, thic father's gens and the mother's narre become important; this is precisсly what we do find in the oldest Pali litcraturc.
Thc usage in question-a singic child with several mothers-is found cxplicitly in vii.2.5.: frürü sisum na mätari rihāne ; i.140,3: tarele abhi mailará Siśum ; and in viii. 99.6. The plural or dual'mothers' in the sense of parcuts is cxcluded, though cvcn this would be highly significant. Panini vi.3.33 : pilarāmātari ca chandasi only shows that the compound could be used in thc dual scnsc, as in Rv. iv.6.7 : na mätaräpilarā, to mcan parents. By itself, matarii as dual would at Icast indicate two mothers, which suffices for our purpose. Whcrc a specific intcrprctation is given (as occasionally by Siyana) we have the parents as the sky and carth : dyādi-pythivi; but both arc feminine and x.64.10) calls the great sky also a mother : ula mítá byhad-dive. The common Sanskrit appellation for ancestors is pitaraḥ, 'fathers', showing how natural patriarchal usage had hccomc. Correspondingly we have thic masculinc 'father sky dyaus-pitā (1.90.7;1.164.33 ctc.) as in Grech, and Latin. Why should this onc god common to all known branches of Aryan mythology appear as a mother so often in thc Rgveda ?
Soma had several mothers: lvām rihanti mätarah (ix.100.7 ; also ix.111.2). In fact he was born of seven mothers, ix.102.4 ; jajñānam sapla mälaraḥ who are sisters, . ix. 86.36 : sapla svasäro abhi mataraḥ Siśum navam jajilinam. These seven mothers are presurnally the seven rivcrs : (1.158.5) nadyo málslamáh ; i. 34.8: sindhubhiḥ saplamalsbhiḥ. The point is that they jointly hcar a singic child while there is no mention at all of the father in spitc of the patriarchal naturc of the society in which thesc hymns were chanted; notc again that the Greck rivers were masculinc. Further, though a river is very uscful to pastoral nomads, the superlative worship in ambitame nadītame devitame sarasvali (ii.42 16,“o most cxcellent of mothers, rivers, goddesses, Sarasvati'') scoms characteristic of the pre-Aryan riparian urban cultures. The connection betwecn. amba mother and ambu or ambhas for watcr is neither fortuitous nos to be cxplained psychoanalytically in this case but a fundamental attitude to be cxpccted among people wiose cntirc civilization owed its birth and its cxistence to the river. The primary sanctity of a river like the Ganges as a clcanser of sin belongs to a later period of Brahminism, though apparcnt cvcn in x.17.10. Thesc river-mothers might be mcant in the famous linc yahu flasya mālara*
*In this phrase, the dusal málará is taken on an night and Uras ini, 342.7 and v.5.6: the sky and carth in the remaining Chy, lyt without internal cvidctice in ix. 102.7 '17 is classical interprration chose its own inconsistcticy, strengther! Loy in.33.5 villich has the plural, along with the adjcctive
mi which is unique in the RV and may therefore indicalc conncction with special Brahmin culis. Further, Sayana gives ildukasya as an alternalive meaning, for flassa even on y, 1.15; vi, 17.7 x 9.4. which make it likely that the origin of the poliras not consuleration is actuallly in the cult of the diuer-mothics, perhaps of two rivers. By i1: clf, jahti ir ur:d in the sense of river. uuite un
whuis werd in the senec of river, quitrunarnbiguously in ii. 35.9 ; iii. 1.4,6,9: 1.72.8.-and even of the Leven rivers,

Page #53

ORIGIN OF BRAHIMIN GOTRAS
73
(v.5.6;x59.8;ix.102.7.;ix.33.5;i.142.7,and vi.17,7 with the added qualification pralne =ancicnt) 'the ncver-resting mothers of truth' (or justicc, rta; but notc that Sūyana on v.12.2 takcs rla to mcan water). The cult of the Mothers did cxist, and was very ancient; if derived from that of thic rivers, one would cxpcct thc process to have taken place among pcoplc who still rctained the matriarchal stamp.
The most intcrcsting fact about such a multiplicity of mothers is its conscqucnt cffcct upon thc child. Thc dcisicd firc, Agni , is also horn of scvcral mothers (x.91.6), specifically the seven blesscd mothers (i.141.2), without an apparcnt father. Wc rcmark parcnthctically that thc fire-drill and thc simplcr firc-plough have only two csscntial components, the 'parcnts' of thc firc gcncrated by their friction, the comparison with human procrcation is so natural that both portions of the araņā arc not gcncrally regarded as mothers. Firc is described in one place as seven-tongucd(iii.6.2), a natural figure of spccch for thc flames. But onc hymn carlicr we have Agni as with scvcn hcads (iii.5.5.): pāti nābhā saplašīrṣānam agniḥ, in onc of his forms at least. Thus it is logical to find that Soma also has scycn faccs or mouths in ix.111.1.
The correspondence of onc hcad per mother can be still better proved from a myth which has been recorded later, namely the birth of Skanda (Mbh. 3.214.). Hc has actually six mothers, thc Plciadcs, whence his name Kārttikcya. But his other namcşanmälura clcarly mcans with six mothers”, and he has six hcads: onc from cach mother as wc arc told cxplicitly in most accounts of his birth. Thc Mahābhārata story is a bit mixed in its details, saying that hic was fathcrcd by Agni who was cnamoured of the scvcn wives of the scvcn ļķis (identificd with components of Ursa Major ; thcsc husbands' arc presumably latcr, sccing that thcy never gain the importance of the Mothers, nor of the collective vcdic gods likc thc Maruts, Rudras, Vasus). Agni's.rcjcctcd wisc svāhā (mcrcly the sacrificial call) then successively assumed the form of six of thcsc scven ladics to couple with thc firc-god; the combined scmcn was pourcd into a lakc to gencratc thc drcad Skanda. Thc duplicated fşi-wives arc cast out on suspicion of unchastity, and adopt Skanda as his mothers. The great Mothers (of thc wholc universe, but scvcn in number) arc asked to kill Skanda, but thcy too adopt him jointly instcad. The story is an obvious cffort to combinc several versions into onc whilc rctaining and cxplaining away thc six mothers with no particular father. Skanda bcing identificd with a form of, or ostcncr as son of, Rudra, we have a still later purā ņic story whercin hc is begotten of the sccd of Siva which Pārvati forces upon Agni in her anger at the interruption; this forms a sort of prcfatory addition to the other story.
10
.

Page #54

D. D. KOSAMBI
Sarasvat is variously given as son or consort* of the river goddess Sarasvati, just as Dakşa is both father and son of Aditi. The confusion, natural conscquence of development from matriarchal cults, suggests the identification of Tvaştp with Tvâştra, at Icast in principle. Gods with scveral hcads would bc associated with the cult of several conflucnt rivers. To continuc: rgvcdic Vişņu has a wife (sumajjānaye vişnave, i.156.2) and several mothers (iii.54.14) while viii.20.3 cquates him to Rudra and the much later Vişnu-smsti (1.56) calls him Saptabirşa without explanation. Both blocks of the firedrill can simultancously be mothers of Agni (v.11.3) Thus Agni or his hcavenly representative the sun (born of heaven and carth) is dvimālā in i.31.2;i.112.4;iii.55.6-7; he is thrcehcaded in 1.146.1 but morc naturally four-cycd in i.31.13 and divisīrṣa in the Sabdacandrikā. The clephant god Gancía is also dvaimītura (Amarakośa 1.1. 140). The BỊhadratha king Jarasamdha was born of two sisters, in two scpa. rate halves later joined together (Mbh.2.16.12-40), which rationalizes the twomother tradition. Rāma cmulatcs Indra and Thraciona in killing a threehcaded demon Tribiras (Raghuvamba 12.47; also Rāmāyana), The Sahdakalpadruma refers to Kālikāpurāna 46 where Hara is called Tryambaka for having been born of thrce mothers. Būhtlingk-Roth give Tribiras as an cpithet of Kubcra (whose thrcc lcgs relate him to the triskelis and the threestrider tripāda Visnu) as well as Siva who in turn is made four-headed in the Tilottamā cpisode (Mbh. 1.203.26) and known both to literature as well as inconography in a five-headed pañcamukha form. Nágas with two, five, seven heads occur in Mbh. 1.52.20, carrying us back to Mesopotamian seals. Even the old Aryan god Varuņa is oncc called four-faced (v.48.5 caluranīka), and again lord of his seven sisters (viii.41.9) thus substituting for somc prc-Aryan dcity ; Indra as saptahá (x.49.8) was too open an enemy (cf.viii.96.16) for this assimilatory treatment. The names Navagva and Dasagva, meaning of nine and ten parts respectively, give clcar indication of ancient Rigvedic groups of ninc or ten pricstly clans of cqual status with the oldest Angirasas (x.62.6; the Nayagvas arc against Indra in i.33.6?). Yet cach is used often in the singular as representing the conjoint group. This could casily arisc from or give risc to the many-headed representation, as for cxample the 'first-born' ten-headed Brāhmaṇa of AV.iv.6.1,or a seven-faced Daśagya Angiras in iv. 51.4. Tvaştr creates BỊhaspati from the essence of cverything (ii.23.17) and also crcates fire (x.2.7 ;x.46.9 ;i1.1.5); but the latter embryo is gencratcd by ten maidens (i.95.2) symbolizing the fingers that twirl the fire-drill, reminiscent of the Vestals. Agni is threc-hcaded and saptarasmi in 1.146.1, just as Tvāştra is in x. 8.8; Brhaspati is saplarasmi and saptāsya in iv. 50.4. As for mother-right, Namuci's army recruited women(v.30.9) to the derision of Indra. The Mothers join Skanda's army (Mbh. Vulgate, Salyaparvan), and have still to hc propitiated by his worshippers. The cow-mother Pęśni is mother of the Maruts, and in
*Qingu, taken as consort by Tiamat after the killing of Apal', scems also to be Tiamal's son (Lange don's translation of the Enuma Elif, ii.31. 1.41), Similarly Tarrauz and Lhtar.

Page #55

ORIGIN OF BRAHMIN GOTRAS
75
viii. 101.15 ‘mother of the Rudras, daughter of the Vasus, sister of the Adityas' yot ncvcr riscs high in the panthcon. In vicw of this rather mixed thcogony, not much can be made of thic phrasc sivas tvasță in thc āpri-hymn v.5.9, for Indra is called śivatama in viii. 96.10. Indra is also ajätasatru (v.34.1 ;viii.93.5), Dhīma in many places, cvcn vişnu in i.61.7 and perhaps rudra in viii. 13.20. That : is, many of the later god-names arc purc adjcctives so that the fact of a god having a good Sanskrit or Indo-Aryan namc docs not necessarily make him a god of the Aryans from his bcginnings. Even the solitary occurrence of lakṣmt (x.21.2) in thc Rgveda is as an adjcctivc.
Thc clumsily patchcd Skanda lcgcnd with its gaping scams is particularly rcvcaling. Without it, we should have assumed, as is done for the modern trimūrti and Dattatrcya, that a multiplc-hcaded god is merely the fusion of that number of malc dcitics, i.c. of thcir cults, Icaving the ancient Brahmi unexplained. But let us first look at the completed patriarchal transformation of such multiple parentage. The introduction of Agni in the Skanda story takcs us only half-way. We have noted that two great gotra-founder rşis with fictitious names, Vasiştha and Agastya (also known as Māna), are born of the combincd sccd of Mitra and Varuna, from a jug or a lotus: two fathers but no mother; this mcthod of gcncration appcars down at least to the siddha Bhartphari, Bharatari or Bhartri of the Kānphäță scct. The essential is the denial of a mother,* these grcat men bcing ayonisambhava, not of woman born. - I suggest that this ingcnious devicc bccamc nccessary because a patriarchal socicty had invaded and conqucrcd by force, but thicse žşis became nevertheless 'originators of gotras. Later the seven sages arc born directly of the four(in some versions cvcn fivc-) hcaded god Brahmi, without female intervention. Yet the names of the 'scven' arc scen to be discordant among the various lists, while the one sage not born of Brahmã at all is kuśika Visvä mitra, thc only truc Aryan gotra-founder. He is rcally a stranger to the seven, even though his book in the Rgveda is pcrmcatcd by Jamadagni influence. Now not only do the scven mothers, the river-goddesses, continuc to hold their high position in thc Rgveda, but the divinc representative of the priesthood, Běhaspati, is scvcral times called scvcn-faced (iv.50.2 ctc ; Sāyaṇa often takes saptāsyas as denoting the Maruts, fathcrcd by Rudra). The conclusion is that a pre-existing matriarchal form of society shows itself through the myth of several mothers jointly giving birth to a god with an equal number of heads or faces. These
* An even better examplc is thc Mandhăts, Icgend. The king is perhaps mentioned in i.112.13, viii.40.2; the word clsewhere in the Rgvcda mcans pious'. In the Mahabharata (3.126) we have his father Yuvanāśva drink cnchanted watcr in Bhrgus's āframa (an inversion of bathing in the enchanted pool), and so become pregnant, the son bcing ultimately born through his side and (in thc vulgate Drohaparvan 62) sucklcd on Indra's finger. This is a complctc repudiation of maternity, as with the couvadc. Mbh.3.127 'has rationalization, by reversal, of the many mothers, Jantu is born jointly of king Somaka's hundred wives, then sacrificed in a yajia, by which cach of the hundred mothers conceives a complctc son. (cf. Kathāsaritsagara 13.57-05). The Southern recension substitutes jyesthāyām samajāyata for strišale samajāyala, rationalizing still furthcr.

Page #56

76
D. D. KOSAMBI
mothers, as representatives of local tribes or gentes, are later replaced by eponymous Brāhmin ancestors, the rsis. Indus valley scals show male animals (single and multiple) which may be interpreted as totemic. The polycephalic god is also prcsent and the civilization has therefore started before and gone beyond the stage of pure worship of his mothers, the rivers or other goddesses. It will be objected that so highly developed a civilization could not have retained matriarchal tradition to such an extent as our analysis requires, but actually there is nothing against it. The main conditions are a relatively undisturbed and rapid advance from the primitive to the urban stage, made possible by the river and its isolat-. ing desert; further, the comparative unimportance of fighting and the warrior in the development of the civilization. Archacology, though incomplete, supports this, whatever the means (naked force, or religion) adopted by that extinct society to preserve internal class divisions; the transformation of the many-headed god into Brhaspati and Brahma suggest religion rather than violence. Even in the epic period, rivers continue to bear heroic sons; the great figure of the Mahabharata war, Bhisma, is born of the Ganges and a human father, Samtanu.
Turn now to Trigiras Tvastra. This personage is supposedly the son of the ancient creator-god Tvastr; a priest-though the father is nowhere called that whence it is a sin to kill him; and in some way an immortal god-priest or else the hymn describing his own killing(x.8) could not have been ascribed to him against all reason by the Anukramani. The father' Tvastr is later enrolled among the Adityas as well as among the Rudras; he share; the adjective visvarupa with his son, but has not three heads. Nothing is said about the mother who bore so remarkable a son, one who is associated with rivers in the form of 'snakes' springing out of his shoulders, as we have seen in Iranian legend. One would guess that he is the son of three mothers, whether also of Tvaste or not. It cannot be a more accident that we find another (nameless) god with three mothers, of whose father there is no mention at all, and who is carly identified with Rudra. This is Tryambaka 'with three mothers', worshipped according to vii.59.12: tryambakam yajamahe sugandhim pustivardhanam. The Taitt. Sam. i. 86 calls Tryambaka Rudra and tells us that his animal is the mole. Later we have Tryambaka translated as 'three-cyed', for which there is no philological support but which does serve to climinate the three mothers; it also explains the three eyes of Rudra-Siva. We have another reference in ii.56.5: ula trimālā vidatheṣu samrāļ, to an unnamed god (probably Agni) who has three mothers and is supreme in the divine assembly; the hymn, it will be recalled, deals with several triple deities. This trimätä is glossed by Sayana as trayāṇām lokānām nirmalā, creator of the three worlds; which, though silly as an explanation, gets rid of the awkward and incomprehensible three mothers while showing that the reference was supposed to be to some high god. The conclusion is again that one branch of culture contributing to the

Page #57

ORIGIN OF BRAHMIN GOTRAS
77
Rgveda had a living tradition in which maternity could be joint and paternity quite unimportant. It is for this reason that Tvästra's severed heads could give names to Brahmin gotras, for they must actually represent matriarchal gentes to begin with. It is not the mother-goddess who has three faces, like Hecate or Artemis among the Grecks, but the son born of three mother goddesses. Just what ancient chain connects our myth to the story of Herakles killing the three-headed Geryon, capturing Kerberos, or decapitating the Hydra we cannot consider here, for we have not as yet enough glyptic evidence from the Indus and Mesopotamian regions.
This can be rounded out by other myths, usually dismissed as trivial but which can now be seen to form connective tissue in the body of vedic mythology. Indra drank the soma by force in Tvastr's house (iii.48.4;iv.18.3) thus presumably thrusting himself upon Tvastr's tribe, or depriving him of power, or both. It is thought by some that the father whom Indra took by the foot and smashed (iv.18.12) is Tvastr himself, but this is highly improbable. Indra's father is nowhere. named, (nor is Indra reported anywhere as asaulting Tvastr) and his mother is doubtful too, though he is enrolled among the growing list of adityas, sons of Aditi. The later aditya par excellence is the sun, while the first is Varuna; both Tvaştṛ and Indra occur in a continously expanding list, and it is not clear that Aditi was a pre-Aryan mother-goddess, being once even cited in the masculine gender. The later Pañcavimsa Brāhmana (xii.5.18-22) reports that Indra suffered from cyc-discasc after killing Vṛtra, and was lulled to sleep by the daughters of Tvastr. These daughters generate fugitive Indra from the cows in which he had hidden himself; parallel versions show that the cows themselves are the daughters of Tvastr, so that the whole story is perhaps one of rebirth from several mothers, i.c. adoption. One may note that Durga is called Tväṣṭi(for Tväṣṭri) in the still later Devipurāņa, and a living cult of Tvastr(or his son ?) scems indicated only by the Paraskara Grhya-sutra ii.15.5. The adoption of Indra by the daughters of his predecessor is meaningless by patriarchal standards; cither Tvastr or his son would have had to adopt the war-god for its validity. What we do see is that not only did Aryans adopt some pre-Aryan Indic gods but assimilation in the opposite direction was also attempted. As for the three heads of Tryambaka becoming three eyes, we have a distant parallel in the Tväṣṭra story. Sat. Brah. iii.1.3.12-17 says that a special eye-ointment from mount Trikakud must be used.Trikakud means with three peaks, points (or even heads). The mountain was the transformation of Vṛtra's eye after that demon had been killed by Indra; but Vṛtra was the demon created by Tvastṛ to avenge his sons's murder by Indra. So the cycle is complete. The variant details of this and other similar narratives show that some background story ich could not be forgotten was adopted by several different people at various times for vcdic purposes; the principle is the same as that of the starred reading in textcriticism, on a different level. It is at least plausible that this faded craftsman

Page #58

78
D. D. KOSAMBI
god Tvaştı who is identificd as crcator with Varuna(iv.42.3) as well as the later Prajāpati, who appcars as a Rudra as well as an Aditya, and who is connected with multiplc-headed gods from Trisiras to Bșhaspati, is not originally an Aryan god with fixed position in the pantheon, but a figure from the preAryan background which could not be suppressed altogether in spitc of his conflict with Indra. .
The three possible mothers of Tribiras could very well be the original of a femalc trícd which occurs repcatcdly in thic Rgveda(i.188.8 ;i.3.8 ;iij.4.8;X. 110.8), Iļā, Sarasvati and Bhāratī. Thc last is thc carth, perhaps herc as a spccial goddess of the Bharatas. Idā is also the mother of Agni (jii.29.3) as personification of the lower wood of the fire-drill. Mosè important of all, she is the mother of Purūravas(x.95.18). Since this Purūravasis virtually thc founder of the lunar linc of kings, we have a complicated sct of puránic lcgcnds making Ilā a son of Manu, but transformed into a woman by stepping
ng into a grove sacred to Pārvati. The original legend had to be twisted, presumably bccausc a line in the patriarchal world cannot be properly founded through a daughter of Manu. We have already seen the prototype of the metamorphosis in the ambivalence of the sky-god or goddess and such changes of sex are far too common. Indra himself (1.51.13; AV vii. 38.2. Sat.Bráh. iii.3.4.18); AsargaPlayogi(Sāyaṇa at the beginning of viii.l, and the Sarvānukramani; Bịhaddcvată vi.41); Nārada, king Bhangásvana(Mbh.13.12, vulgate) and the 'monkcy' Riksarajas (in a probably apocryphal addition to tic Rāmāyaṇa) after hathing in enchanted pools; Sikhandlin who killed Bhīşma(originally and significantly named Ambā in a previous birth) all change sex, and somctimes both bcget and bear children. The roots go very far back, for the Tiraści of viii. 95.4 is the scer of the hymn, but the name is femininc in dcclension and masculine in usage. The grove and particularly tlic pool which cffcct the metamorphosis (which will be found cvcn in a tale of the Arabian Nights, and thc Qissah Halim Tat) has somctimcs been equated to the fountain of youth, as with the rejuvenating immcrsion of Cyavana*. The actual transformation in the first instance being from malc to femalc, thcy are much more likely to rcpresent places dedicated to the mysteries and initiation rites connected with the cult of one or more mother-goddesses----places which men could enter only to cmcrgc cmasculated, performing thercaster the functions of women, presumably in the service of the goddess. Some such prc-patriarchal initiation must bc the proper cxplanation of the verscs at the end of viii.33, particularly 19: strī hi brahmā babhúvitha 'thou, O priest, art become a woman.'
The forcgoing, I believe, will suffice to show how correct and useful a guide Engels's "Origin Of the Family, Private Property And The State" has
Cuayana clory is not a parallel at all, for therejuvenation is performed by the ASvins and the immerhorepetir later (Mbh, 3, 123.15-17) than the Ryvedir, whicre the sage regains his
dinira hai kiedrawn off like a garment (7.74.5; 1,116,10). This is the older version, based uptoprimitive wonder al pake cacting off his skin to appear rejuvenated.
T

Page #59
_