American Oriental Society http://www.jstor.org/stable/604266 .

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

.

American Oriental Society is collaborating with JSTOR to digitize, preserve and extend access to Journal of the American Oriental Society.

http://www.jstor.org

THE YOUNG WIFE AND HER HUSBAND'S BROTHER: RGVEDA 10.40.2 AND 10.85.44
M. B.
EMENEAU

and B. A. VAN NOOTEN

UNIVERSITY OF CALIFORNIA, BERKELEY

Textual problems are presented by devikdmd in the marriage verses RV 10.85.44 and AV 14.2.1718; problems of interpretation are presented by these verses and also by the occurrence of devf- in RV 10.40.2. These problems we hope we have solved, or at least eased, by consideration of the social structure involved in the niyoga institution described in the smrti texts, the similar custom seen in many communities in present-day North India, and the related 'jesting relationship' described for roughly the same communities (and others). Chronological problems are discussed, especially the establishment of a chronology for the semantic development of devi-/devara- from 'husband's brother' to 'husband's younger brother'.

THERE HAS BEEN A CONSIDERABLEamount of scholarly discussion over a period measurable in millennia about the terms devr'- and niyoga- that are involved in the Vedic verses of the title and in their interpretation. Our handbooks read, on the whole, as if the problems had been solved, or at least they provide statements and definitions that sound as if they were solutions. However, problems still remain. Towards the solution of some of them there is some evidence that has hardly been used so far; not all gaps in the evidence can be filled until more information surfaces. One of the basic problems is the reading of the Rgveda (RV) passage 10.85.44. Our usually accepted text of this verse in the marriage hymn, addressed to the bride, reads in pada c as follows: vfrasutr devakamd syoni (metrically: vTrasiuur devdkamd siyona; with verb edhi from pada a) '(be) hero-bearing, loving the gods, pleasant!' The second word, devakamd 'loving the gods', is found in all the published text editions, beginning with Max Muller's editio princeps with Sayana's commentary. This was printed in six volumes appearing from 1849 to 1874, the last volume (1874) containing our hymn; the whole RVtext, without the commentary, had been reprinted in 1873, one year before the last volume of the editio princeps appeared. But the appearance of this edition was punctuated by Aufrecht's edition of the text (without the commentary) in Roman transcription in two volumes in 1861 and 1863; this edition also reads devikamd. All early complete translations of the text, from the unsatisfactory Langlois (1848-51; n.b., long before a printed text was avail-

able)' to Wilson (1850-88; the last volume, with our hymn, is much later than the corresponding printed text), show that they are based on the vulgate reading. Similarly, Grassmann's translation (1876-77), which is based primarily on Aufrecht's text, reflects this reading, as does his accompanying RV dictionary (1873). Sayana's commentary, as printed in MUller's and later editions, quotes devaikamd but gives no gloss or explanation. Why insist on this history of the text? There are two reasons. First, our basic dictionary (PW), that published by Otto Bohtlingk and Rudolph Roth, the latter being responsible for the Vedas, fails to list this RV passage as authority for devakdma-, though it lists the few other passages in which the word occurs. Our passage it lists under devtkama- 'loving/fond of brother(s)-inlaw'. The relevant section of PW was dated 1859-61 and was published in 1861, long before the relevant section of the Muller text. Our second reason is that the only other text that Roth lists in the dictionary entry as authority for devtkdma- is the Atharvaveda (AV) passage 14.2.18, which is parallel to our RVpassage. The AVpassage is indeed parallel to that of R V, but it by no means completely agrees with it verbally, and in fact it is somewhat inflated as compared with it, and contains devikama- in two successive verses, 14.2.17, 18.
I We are indebted to Stanley Insler, Professor of Sanskrit, Yale University, for information on this translation, which was not available to us.

481

482

Journal of the American Oriental Society 111.3 (1991) (1905), who had written (as above): "devak- . . . which is therefore more probably the true reading [of AV]." That Ppp. really read devrk'in both verses is authenticated by the manuscripts of this school discovered in Orissa in modern times.' On the other hand, the later edition of the AV vulgate by Vishva Bandhu (1961), who collated additional manuscripts, reads in both verses devik' without indicating the readings of his manuscripts. Whitney-Lanman had been following the usual doctrine of modern Western Vedic scholars when judging textual matters: when there are Vedic textual variants, the presumption is that an R V reading is the more original and to be preferred.3 It should be noted here that, when Lindenau in 1924 published a "zweite verbesserte Auflage" of Roth-Whitney's AVtext of 185556, he emended their text of our verses to read devakW, following the RV, as did Whitney-Lanman in their 1905 translation (as quoted above). Not all scholars, however, have been equally firm in holding to this assumption. On the variants with which we are concerned Alfred Ludwig, as so often,4 preferred the AV reading (devik') to that of the RV vulgate (devak?) when he discussed the RV reading. On the other hand, Oldenberg in his very influential Prolegomena (1888: 324, fn. 2) and Noten (1912: 289), though obviously troubled by the problem, preferred the RV reading on the basis of better attestation (in grhyasiitra texts and the Samaveda), but then baldly (1912) "da es im Rv. steht." He attempted in 1888 to explain the AV reading as due to another occurrence of devr- in the neighborhood, but in 1912 abandoned this (tacitly), mentioned a different, neighboring occurrence of devr- in the RV (but not the AV) text, and finally, saying that the verse
' There has been no publication yet of the pertinent passage, but we can report the readings owing to the great kindness of Michael Witzel, Professor of Sanskrit, Harvard University, who has supplied us with an electronic version of his researches on the Orissa manuscripts of the Paippaldda school text. 3 So for the readings in these passages in Bloomfield and Edgerton 1932:296, ?633, though with the reservation:"Either reading would however make sense." 4 See Windisch 1920: 374: "In der ersteren erbrtert er den Wert, den die Varianten der 0brigen Veden, besonders des Sdma- und des Atharvaveda, gegendber dem Texte des Rgveda haben." He refers there ("In der ersteren") to Ludwig's 1889 article; on p. 51 Ludwig compares the R V and AV readings that we are concerned with. Ludwig (1876-88) in vol. 5 (1883), p. 400, on RV 10.85.44 says explicitly: "devakkmA wol eine ist corruption."

Roth's reading in PWfor the RVpassage is discussed later. The AV text must be dealt with first; most of its variants from the RV text are irrelevant in this connection and are ignored here. The AV text was published in 1856, long before the final volume of the RV (that containing our RV passage) published by Muller in 1873-74 or Aufrecht's second volume in 1863. The AV edition was by Roth and his pupil W. D. Whitney; this edition contained the reading devik? that we are interested in. After the AV text publication came the PW dictionary entry in 1861. It is to be noted, for completeness, that the fourteenth book of AV was treated, textually and comparatively, in 1862 by E. Haas and Albrecht Weber; there is nothing in that publication pertinent to our discussion. And, again for completeness, there is no Sdyana commentary for this book of AV. The textual problems of the AVverses were treated in detail in the AV translation with textual and other commentary published by Whitney and his pupil, Charles R. Lanman, in 1905. The pertinent word of RV's pada c appears similarly in pada c in both AV verses, with the difference in reading (RV deva-: AV devr-) already noted. However, Whitney-Lanman record that the AV manuscripts in both verses have variant readings, devrkl?of the vulgate, i.e., the 1856 edition, and devak? agreeing with RV, "the majority. . having, with RV and Ppp., the latter [i.e., devak], which is therefore more probably the true reading." The statement that "Ppp." has a reading like that of RV is in fact oversimple and in part emendation. The Paippalada (Ppp.) school version of AV was at that time known only as a very corrupt text in a manuscript from Kashmir. It had been published in facsimile in 1901 by Bloomfield and Garbe, whence the Whitney-Lanman readings. Beginning in 1905 and ending in 1938, L. C. Barret published book by book (except for book 6, which Edgerton published in 1915) in a series of articles in JA OS, the text both as it appears in the manuscript and as he reconstructed what was intended. At times, as here, his reconstructed text contained what can only be called emendations. The Ppp. reference for its verses corresponding to AV 14.2.17, 18 is AV Ppp. 18.8.8, 9 (Barret 1938, esp. pp. 583-84). Padas c of verses 8 and 9 read in the corrupt Ppp. text respectively: prajavatT devakdmd syond. BarvTrasaur devrkamd, and vTraseir ret's reconstruction contains in both verses devakdmd; in a note on verse 8 he has: "With this and the next cf RV 10.85.44; ApMB 1.1.4; and HG 1.20.2: these confirm devakama." I.e., he has abandoned the alternative devrk , the reading of our AV vulgate, in favor of that of the R V vulgate and the grhyastitra texts that follow the RV reading. He is following Whitney-Lanman

EMENEAUand VAN NOOTEN: The Young Wife and Her Husband's Brother

483

(in RV) is a general wish for prosperity ("Gedeihen"), ends with the statement that devaik is "bestens am Platz" as in the other RV passages where it occurs.5 We are left with the problem of why Roth in PW records the R V as reading devrk. Considerable search has failed to discover that he ever discussed the reading. Alternative solutions are:
(1) Some of his manuscripts (he was not working from a printed edition; see the chronology above) had this reading, and, with or without comparison with AV, he judged it superior; (2) Without having any R V manuscript variants, he, like Ludwig later, judged the AV vulgate reading superior to the RV ulgate reading, and in effect emended the R V text; (3) His PW recording is an error.

RV 10.40.2 is very much to the point. kiiha svid dosa kfiha vitstor aivina kiihabhipitvam karatah kuhosatuh k6 vam vidhaveva devdram rayutra maryam na yosa krnute sadhdstha a Where are the Aivins in the evening, where in the morning? Where do they enter for lodging [abhipitvam], where have they spent the night? Who puts you [Agvins] to bed like a widow her husband's brother, like a young woman a young man, in a habitation?7

This passage is not without its difficulties. The construction gayutrd . . . (object) . . . Vkr (middle voice) is a a7aE kEyo6pEvov the language, as is for all practical in

Alternative (3) seems most implausible, considering how careful a scholar Roth was. As between (1) and (2), only reexamination of Roth's manuscripts could decide. Presumably the manuscripts are still available (at London, Oxford, Paris), but it is a task we cannot undertake and must leave to those who have easy access to them.6 Up to this point we have treated the matter as purely a textual problem. There is however another approach which should allow some advance. Why is the bride's prospective brother-in-law singled out for conspicuous mention? In other words, what feature of social structure is the background for the verse? Vedic literature contains several other instances of the word devr- which should be examined. There are two other RV passages which contain the word. In RV 10.85.46 (two verses after our passage) the bride is to be samranji 'queen, woman superior in rank' over her father-in-law, her mother-in-law, her husband's sister, her husband's brothers (devkrsu). This is too allinclusive to be of any use to us.

purposes the word sayutrd.8 But the intent is clear-a widow sleeps, i.e., has sexual intercourse, with her (dead) husband's brother. This is not to be interpreted in terms of the comparatively unstructured sexual mores of the late 20thcentury sexual revolution in the Western world, where in fact the situation envisaged here would probably have incestuous overtones. Among its several wellstructured sexual patterns, India has had and has one which makes it necessary or desirable that a (young) widow should take as sexual partner, either with or without some further marriage ceremony, her dead husband's brother. This, we suggest, was already the custom underlying the phraseology in RV 10.40.2. In modern North India in many caste communities it is specifically a younger brother of the dead husband that is involved (as we shall see below). This is not, however, Sayana's interpretation (see fn. 8); he glosses as devararm bhartrbhrdtaram'husband's brother'. There is need of investigation (undertaken below) into the chronology of this change of meaning for devara-. Moreover, a 'jesting relationship' has been identified in some communities between a young wife and her
7 This is Burrow's translation (1973: 100), where the interpretation of abhipitvwm is at issue. He translates devaram too specifically as "husband's younger brother"; see discussion below. 8 For this adverb 'on the bed' (Sayana's sayane), see Burrow 1955: 131, 279. Interpretation of what seems to be sayutrn in RV 1.117.12 is much disputed. For pada c, Geldner 1951 translates: "Wer nimmt euch zu sich ins Bett wie die Witwe den Schwager . . . ?" This gives full value to the middle voice of the verb (krnute). Sayana seems a little mealy-mouthed in his gloss: sayutrd sayane vidhave 'va yathd mrtabhartrkd ndrT devaram bhartrbhrataram abhimukhlkaroti

5 The other RV occurrences of the word are 2.3.9, 3.4.9, 4.25.1, 10.42.9, 10.160.3. In all these the word has specific or general reference to a god or gods. In our passage that reference is lacking and the word is merely a hoped-for characteristic of the wife. Geldner's translation (1951) has no textual discussion of the verse; his footnote merely refers to its use in the Sdnkhayanagrhyasuitra. 6 Delbruck (1889: 138/516) in his study of Indo-European kinship terms has: "So ist wohl Vers 44 [i.e., RV 10.85.44] mit Bbhtlingk-Roth devfkama 'die Schwdger liebend' zu lesen...." Is he opting for alternative (1) or (2)? Or is he following Ludwig?

484

Journal of the American Oriental Society 111.3 (1991) varah kasmdt/dvitryo vara ucyate 'From what is devara derived? (He is) so called (because) he is the second husband'. The editor, Lakshman Sarup (1921: 48-49, fn. 5), for sufficient textual reason, finds this to be an interpolation, and (1927: preface, pp. 19-20), since it is not in the text as given by the commentator Durga, it must be later than (about) the thirteenth century A.D., the period given for Durga. Leaving behind Vedic texts of all sorts and coming down a time-span of at least half-a-millennium, we find conspicuous mention of the devara in the law books. Most conspicuous are the dharmasastrins'statements about and discussions of an institution called niyoga. This appears most fully and authoritatively in a passage in the Manusmrti (9.56-63; text, Jolly 1887; translation, Biihler 1886). The passage has a selfcontradiction, and leaves some things ambiguous and open to interpretation. Discussion has gone on for the approximately two millennia down to the present; it has at times been all too easy to present arguments from extra-Indic analogies and to find, e.g., that the niyoga is parallel to Western and Judaic levirates. The niyoga in the Manusmrti was not a marriage.12 It took place when on the death of a woman's husband there was no issue, i.e., presumably male issue (vs. 59 samtdnasya pariksaye 'on destruction of continuity of the lineage'; vss. 56, 58 dpadi'in misfortune', i.e., failure of male issue). Vss. 57-58 are in effect a preface, stating (vs. 57) that the wives of brothers are forbidden to the brothers-in-law except under such circumstances, and (vs. 58): jyestho yavlyaso bhdryam yaviydn vi 'grajastriyam patitau bhavato gatvi niyuktdv apy andpadi An elder (brother) having gone to the wife of a younger or a younger to the wife of an elder, even though they were appointed (in niyoga), both become fallen (from dharma, outcaste), except in circumstances of misfortune (i.e, failure of issue).

brother-in-law (younger in modern times); this is discussed below and plausibly provides the clue to the use of devwkamd in the marriage hymn of AV and (possibly) RV. Several passages in the veddnga texts (i.e., the interpretative literature on the Vedas) have devara- (the classical equivalent of, derived from, devr'-)as the prime character in the situation. In the isvaldyanagrhyasuitra 4.2.18 (Oldenberg 1886: 239; Kane 1941: 618), the widow, who has lain beside her dead husband's corpse on the funeral pyre, is made to rise up from the pyre by recitation of RV 10.18.8 by one from a list of several men, of whom the brother-inlaw is named first:devarahpatisthdpanryo 'ntevdsijaradddso vd 'her brother-in-law, being a representative of her husband,9 or a pupil of her husband, or an aged servant'. Similarly, in other passages (e.g., Rgvidhdna 3.8.4, quoted by Kane 1941: 618) dealing with the same situation, there is mention specifically of the devara. The Brhaddevatd, later than the Nirukta but probably pre-Paninian, in 7.13 has the RVverse 10.18.8 spoken by a younger brother-in-law, but without the word devara-: bhrdtd kanrydnpretasya nigadya pratisedhati 'the younger brother of the dead man, reciting (the verse), prohibits (her from ascending the funeral pyre)'. The following verse (7.14) has the word devara- (devaro na bhaved yadi 'if there should not be a brother-inlaw'), but the ritual specifying of the younger brotherin-law in the previous verse does not require that devara- in vs. 14 be so restricted in meaning.'0 One of the vedangas, Yaska's Nirukta, concerned with etymologies, at 3.15 quotes RV 10.40.2 and etymologizes vidhavd- and devara- (n.b., the classical form of the word). This text is dated, with the usual query, to the seventh century B.C., or, at least, prior to Panini. It gives the usual Hindu derivation of devr'-/devarafrom Vdiv 'to play"'-but also has the passage: de9 A variant interpretation, which takes the four nouns to be four alternatives ('or a repr?. . . '), seems on the whole less probable, but for our purpose the variation is irrelevant. 10 Macdonell 1904 in his critical notes records that vs. 7.14 is found only in the longer recension, that he demonstrates to be more original than the shorter recension, which is an abridgement. This matter does not affect the point in which we are interested. On p. xxiii, he places the work in the fifth century B.C. 1 Since Vdiv 'to play' is without an Indo-European etymology, this Hindu grammarians'etymology for devr'-is of no value for this noun inherited from Indo-European. Szemerenyi 1977: 87-88 gives the list of Indo-European cognates for devrand the unsatisfactory etymological origins that had been

The appointment/ authorization, made by the husband's elders (guru-), is niyoga- (the word is used in vss. 61, 62, and 65; otherwise forms of the verb ni-yuj-; the man is niyukta-, the woman niyuktd). The appointed duty is stated in vs. 60: suggested previously. On pp. 177-78 he deals with further, implausible etymologizing that has accompanied anthropological theorizing about the Indo-European kinship system as a whole. 12 As is suggested, e.g., in the definition in Monier-Williams' dictionary (1899).

EMENEAUand VAN NOOTEN:The Young Wife and Her Husband's

Brother

485

tu vidhavayam niyuktas ghrtakto vagyatonigi na ekamutpadayet putram dvitiyamkathamcana to withghee, He who is appointed thewidow,anointed silent, at night should beget one son, by no meansa second. But some (vs. 61) allow the begetting of two sons, presumably in order to secure the succession (and considering the high level of infant mortality). Vs. 63 makes it clear that they shall not act from love (kdmatah). The man appointed (vs. 59) is the woman's devara (the brother of her [dead] husband) or, presumably failing such a person, a sapinda of her husband. The word devara- in vs. 59 must be defined as 'woman's husband's brother' (whether elder or younger than the husband); it does not mean her younger brother-in-law, as it came to mean later. This then is niyoga, but in the following verses (6468), without any transition, the institution of niyoga is flatly forbidden: nd 'nyasmin vidhavananr niyoktavyd dviJatibhih hi anytasmin niyuiijana dharmam hanyuhsanatanam (vs. 64) A widow must not be appointedto anotherman by the twice-born.For they who appointher to another manwoulddestroyeternalrighteousness. In vs. 65 it is said that in the sacred texts having to do with marriage (udvahikesu mantresu) there is no mention of (na krrtyate) niyoga13 (nor is there mention of remarriage of widows). The remaining verses (66-68) state that this beastly custom (pasudharma-) of niyoga which the learned twice-born reprehend (66ab: ayam dvqair hi vidvadbhih pasudharmo vigarhitah) was favored by a king Vena-who is apparently nowhere else, in epic or purana, associated with niyoga. This whole passage (vss. 56-68) has been the subject of much learned, legalistic discussion right down to the present, as has most of the rest of the chapter, which seems to be full of mutually irreconcilable, or almost irreconcilable, doctrines. It may be noted, however, that if a man dies after he is betrothed to a girl (vs. 69) or after he has given a nuptial fee (sulka) for a girl

(vs. 97), the man's brother (devara) should "marry"14 the girl-but in each case there seems to follow a statement that flatly forbids the practice (vss. 71, 9899), just as there was after the description of niyoga. An easily accessible and fairly recent presentation of the discussions on niyoga is that by Kane (1941: 599607), from which the following excerpts are made. Many commentators, as seemingly the Manu passage itself (vss. 66-68), interpret the conflicting viewpoints as a quasi-historical statement: niyoga was once an acceptable rule of procedure, but is not so now. Others have thought of an option; but if that were so, it is hard to understand the second part of the Manu statement (vss. 64-68). The very authoritative commentator Visvariipa (8-9th century A.D.) held that niyoga was the practice of sudras (Kane 1941: 604) and also of ksatriyas, when there was no male to succeed in a royal family (Kane: because of wholesale slaughter in war), in which latter case only a brahman was to be niyukta for the widow. It has been pointed out in favor of this view (with reference to niyoga among sudras) that the Manu text itself perhaps points in that direction by its use of dvijdti 'twice-born' in vs. 64 and (one might add) of dvija 'idem' in vs. 66. Kane (1946: 931) discusses shortly a further point. Some commentators have, it appears, been of the opinion that only a brother-in-law younger than the dead husband could be appointed in niyoga, relying on Vijfianesvara's Mitaksard commentary15on Ydjniavalkyasmrti (1.68), "where 'devara' is paraphrased as 'kanlyan bhrata'." This is refuted by others, defining devara- thus: devaras tasya jyesthah kanistho vd 'his elder or younger brother'. The relevance of the Mitaksard change in the meaning of devara is discussed below. Another smrti text of importance, the Ndradasmrti (Jolly 1885 and 1889), parallels closely the Manusmrti in its description of niyoga (12.80-88). It details the same restrictions and procedures, e.g., the need for continuation of the lineage (vs. 83, samtandrtham), discontinuance of intercourse after this has been achieved

'3 Buhler 1886: 339, fn. 65, says that the ninth century commentator Medhatithi points out that RV 10.40.2 "mentions" niyoga. Since all that that verse says is that the widow goes to bed with her brother-in-law, Medhdtithi could well be right, but the verse does not use the term niyoga.

14 So Buhler 1886 for both verses, but in vs. 97 the form is praddtavyd and in vs. 69 vindeta. Is it possible that the passages mean no more than that he should take the girl (by niyoga)? Certainly, vs. 70's adhigamya, which Buhler translates "espouse," means no more than 'approach sexually', and that verse contains the restriction laid down previously for the niyoga, viz., that there should be intercourse only until there is offspring (dprasavdt). '5 Dated I Ith century by Keith 1920: 447, and A.D. 1120-25 by Derrett 1973: 55.

486

Journal of the American Oriental Society 111.3 (1991) The Kautilhydrthasdstra (Kangle 1963) contains several sections which cover much the same topics on marriage as Manu. In spite of the efforts of Prabhati Mukherjee (1963) to set out the parallels between the two texts, it must be noted that though the Arthasdstra contains much more detail than Manu, it contains somewhat less detail about widows, and, in fact, of the niyoga terminology it has only niyukta- (3.7.6) without definition of the 'appointed' man. In 3.4.37-38, if a woman is widowed (or her husband has gone on a long journey or has become a wandering monk), she may go to (gacchet; marriage is not specified until sentence 40 where vedana- 'marriage' and vettr- 'one who marries' are used) her husband's co-uterine brother (patisodaryam). In 3.4.39 (bahusu pratydsannam dhdrmikam bharmasamartham kanistham abhdryam vd) what has been interpreted as a descending order of preference (sentence 41, krama-) is given if there are many brothers (bahusu): 'one who is proximate [in age], one who is pious, one who is capable of maintaining her, or the youngest if without a wife'. There has been disagreement among the commentators and translators as to whether the last phrase, kanistham abhdryam, intends one person or two; we have followed Kangle (1963: 240) in settling for one. But it should be noted that the whole passage, like that in Manu, makes no firm decision between the woman's brothers-in-law on the basis of their age; nor does this text use the term devara- at all. The epics have material that is pertinent to our enquiry and of an age roughly contemporary with Manu and the Arthasdstra. The Mahdbhdrata (Mbh.) in its books ?dnti (12) and Anusdsana (13) has several passages that use phraseology reminiscent of and obviously related to the statements in Manu (in fact Manu is mentioned as an authority several times, e.g., 13.44.17 and 22). Parallel to the Manu passage on niyoga, if the husband dies after sulka is given and before the marriage is consummated, Mbh. (1966) 13.44.50ab has: devaram praviset kanyd tapyed vd pi mahat tapah '(if the husband dies before the consummation of the marriage), the girl [Ray: virgin-widow] may have sexual intercourse with her husband's brother (devaram) or may undertake great penance." A simile in which niyoga is used is found in Mbh. (1961) 12.73.12: patyabhave yatha stri hi devaram kurute patim anantaryat tatha ksatram prthivTkurute patim As a woman in the absence of her husband makes her husband's brother (devaram) to be her husband, so the earth in immediate succession (to the brahman) makes the ksatriya to be her (the earth's) husband.

(vs. 81), preparation for intercourse by anointing with ghee (vs. 82), the devara may be the brother-in-law older or younger than the husband (vss. 85-86), and, most insistently, that there is to be no "love" between the widow and her devara (vss. 83, 88). But this text, unlike the Manusmrti, prescribes niyoga and does not include in its exposition the Manusmrti's statement (vss. 64, 66) that the 'twice-born' (dvijdti-, dvija-) must not practise this custom. We would conclude that in fact, at the period of the smrti texts, niyoga was practised, when necessary for the continuation of the lineage, in all castes; that there was beginning to be held the view, which later became the only really respectable one for upper castes, that remarriage of widows was reprehensible; and that the compilers of the Manusmrti (perhaps at different periods) included reports of both views, failing to reconcile them, though inclining towards what became the more respectable one and seeming to deny flatly the validity of the niyoga practice. The somewhat more realistic attitude of such a text as the Ndradasmrti was on the whole neglected by later commentators (down to the present) and led to much misplaced ingenuity of interpretation. The very modern writer on law in India, Derrett (1978), in a summary treatment of niyoga, wrote (some sentences are excerpted): "Each caste has its own preferential duties" (p. 51); "[The Rishis] extrapolated leading principles and taught what, in their view, was best, and at the same time left room for usages which were tolerable or barely tolerable. Their patrons and their pupils could not be affronted by direct attacks on certain cherished customs.... [They] are found to ack-

nowledge as existing and worthy of regulation a few institutions which affronted their refined moral senses. The best example is niyoga, whereby a childless widow could be 'authorised' to have intercourse with her husband's younger brother to bear a child.... In due course centuries of contempt for the very practical institution of niyoga led to its disappearance for almost all the higher castes for whom the Rishis' precepts had a great prestige value. And so it could be said-though one wonders how accurately-that by the time of mediaeval commentaries niyoga was obsolete" (p. 52).16

16 There is here the unwarranted specification of devara as 'husband's YOUNGER brother'. The very authoritative study of Indian kinship in general by Trautmann 1986, though it has much on the Indian classical statements, has on niyoga one sentence (p. 247) which in effect translates Manu 9.59, with the same unjustified equivalent 'husband's younger brother' for devara.

EMENEAU

and

VAN NOOTEN:

The Young Wife and Her Husband's Brother

487

This verse is found with slight variation also in Mbh. (1966) 13.8.21.17 In a number of passages in book 1 (Mahabhdrata 1933), when a widow is childless and the lineage is to be carried on (Mbh. 1.111 dpadi, 1.97.10 1.30 samtdndya kulasya nah; note the parallelism with the phraseology of the smrti texts), someone is appointed for the purpose; the phraseology that we have met in the Manusmrti is used-niyoga- for the appointment (1.97.10, 1.111.36) and verb forms (1.97.7, 1.99.15).8 The much-discussed'9 relationship of the Pand.avas and DraupadTas an example of niyoga is hardly simply that; some of the details need to be discussed below. Certainly, the Mahabhdrata general material parallels that part of the Manusmrti which describes niyoga. On the other hand, the Ramayana is a close illustration of the Manusmrti's denial of the validity of niyoga and is a statement of Manu's opposing view that a woman may be married and have a husband once only. Even the slightest suspicion that Slta might have had relations with any other than Rama results in her disgrace. There has, to be sure, been discussion that the relationship between Sita and the two brothers, Rama and the younger Laksmana, might be a thoroughly disguised and rejected instance of niyoga; we prefer not to become embroiled in this difficult discussion, but only to mention that even Indian scholars have seriously engaged in it. If we skip nearly two millennia from the smrti texts, the Arthasastra, and the epics to the present, we find in the ethnologist Irawati Karve's Kinship Organization in India (1965) a wealth of pertinent material on the occurrence of niyoga (a term which she does not use, generally calling it "levirate") among many castes of North India. The book is unfortunately without an index and is hard to use; moreover, many of the statements are made without exact reference to the caste or castes in which the reported phenomena occur. Even so, the greater part of the data is drawn from fieldwork done by the author herself, whose entree and linguistic rapport were infinitely better than those of most previous ethnological reporters. For her the "northern"

17Ray's translation (1883-96) in these passages renders devara- as "husband's younger brother," but the contexts and the commentators quoted by the critical edition on 13.44.50 do not specify age, and so far there is no warrant for so interpreting devara- in the epic. 18 Mbh. (Bombay edition) 1.120.35 is referred to as having devara-, but the critical edition (1933) 1.111.30 makes it clear that this is a corrupt reading in a few northern manuscripts which the commentator Nilakantha followed. '9 E.g., in Winternitz 1897.

and the "central" zones of India together included all the Indo-Aryan areas of North India, plus a few of the tribal communities bordering on these areas. Typical of her statements are: "the custom of junior levirate [i.e., sexual relations between a widow and her husband's younger brother] is found among many castes in the north. Generally Brahmins, Khatri, Kayastha and Vaishya deny the existence of this custom among them but a lady medical practitioner living in the Delhi region for many years assured me that the custom is very wide-spread in villages and is found among all castes [in the context she seems to mean 'all castes' including those previously named]. A widow simply goes and lives with her younger sister-in-law (husband's younger brother's wife) as her co-wife. She is referred to as 'sister' sitting in the house (gharbathelT)" (pp. 131f.); "The custom of levirate by which the widow either lives with or marries the younger brother of her husband is found among the lower castes of Gujarat. Folk-tales, proverbs and songs bear ample testimony to this custom though people get angry at such an enquiry" (p. 173; see also pp. 19-20); "Levirate is found among northern Kunbis20and other castes (Tirole, Govari, etc.), but is not tolerated in the rest of Maharashtra" (p. 178); "[in Orissa] junior levirate is found among all poorer classes. The Brahmins, Karans and Khandayats do not allow such practices, but a more detailed study of families and folk-literature might reveal the existence of the custom among some at least of the higher castes" (p. 193). There is recording of cases of the "junior levirate" among the Gonds of Mahakosal (Karve 1965: 252, n. 10). This is explicit recording of the niyoga institution, with or without marriage, and restricted to relations between a widow and her husband's younger brother, in Indo-Aryan India at the present day, in a wide range of castes below the very highest (i.e., below the dvijas, twice-born). Two millennia ago it was known, but without restriction to the younger or youngest among the brothers, to the writers of the dharmasastra, with no castes mentioned except the disapproving twiceborn. If we are allowed to project the same institution back another millennium (and a half?), we find a very satisfactory social context for RV 10.40.2, in which the widow goes to bed with her devi-. So it was interpreted by, e.g., Geldner (1951) when, without reference to anything later than the grhyasutras, he made a comparison with the levirate; Medhatithi in the 9th century A.D.

20 The Marathas and the Kunbis form about 40 percent of the population of Maharashtra, a fighting and ruling class who call themselves ksatriyas (p. 175).

488

Journal of the American Oriental Society 111.3 (1991) relationship covers a range of behavior from free conversation through sexual jesting to actual sexual relations, it is obvious that the label is not really revealing, but no better one comes to mind. It is also fairly obvious that in both north and south the persons involved in this relationship are potential sexual mates. Above we found that the widow who went to bed with her devi- in RV 10.40.2 was very satisfactorily explained by the niyoga of the law texts, which involves her dead husband's brothers, and the equivalent modern practice involving her dead husband's younger brother. We are now inclined to use as an analogue the modern joking relationship of bride/young wife and her husband's younger brother (devara-) and the linkage between this joking relationship and her sexual fate when widowed, and thus to explain the three (or threeand-a-half) millennia earlier word devfkdmd of the disputed reading in the wedding verse. When she is being married, she is being told, not 'to love (or make love with)' her husband's brother, but 'to be on good terms with/be fond of' her husband's brother, whom she may indeed live sexually with, or marry (if she is widowed). The word is truly a lectio difficilior compared with the bland devakdmd'pious'. We might add too that brahman transmitters of the Vedic texts may well have disapproved of a word that smacked of niyogatype practices, which are disapproved by the dvijdti compilers of such a law book as the Manusmrti. Then, as between the two readings of the AV passage, we would prefer the reading devtkdmd. Given any manuscript evidence for Roth's PWreading devikdmd in R V 10.85.44, we should be inclined to adopt that reading for R V too, or even to adopt it (as did Ludwig), if Roth really produced an emendation on the basis of the AV reading. Modern scholarship concerned with the Mahdbhdrata has investigated the possibility that the marriage relationship between the five Pandava brothers and their common wife Draupadi might be explicable in terms of the niyoga institution as described in Manu (and there are otherwise, as we saw above, traces of a knowledge of the niyoga in this epic), or in terms of the modern North Indian r6le of the husband's younger brother as regards that elder brother's wife. Unfortunately, it seems that there can be no straightforward relationship of either sort. Whatever parallelism there is shows so much distortion in comparison with both the known institutions that for our purposes this matter may be neglected. Certainly, Draupadi's relationship

had referred to this verse in his commentary on niyoga in Manu 9.65 (see fn. 13). Another facet of the relationship between a bride and her husband's brother, specifically younger brother, is described at some length in Karve's book (1953/1965). She describes in some detail for North India's joint family the pitiable position of a young bride in her husband's family. She is torn from her own family of birth and is received with almost completely general suspicion and hostility into a family of strangers, even her husband being a complete stranger to her. Of her husband's parents (and whatever other members of their generation may live in this new family), all are for her objects of deep respect and avoidance, as are also the elder brothers of her husband ("A woman must stand up and cover her head and face if she is in the same room as her parents-in-law, and the brothers ... of the parents-in-law.... Except on ceremonial occasions, she must never be in the same room as her father-in-law or the elder brothers .. . of her husband" [p. 135]). Of the women in her new family, her motherin-law is almost certain to be a domestic tyrant in relation to all the young wives in the family. The younger women of the family, whetherunmarrieddaughters or wives of her husband's brothers, are all suspicious of her and banded together to see that she keeps her place ("There is a traditional 'enmity' between a woman and her husband's sister or [husband's] brother's wife" [p. 137]). But she has one friend, her husband's younger brother (or brothers?): "[in the joint family of North India, in general] towards the younger brother of her husband her behaviour is more free and she may joke with him" (p. 136); "[in Gujarat and Kathiawar, in all castes, there is] a custom which allow[s] a woman to joke and speak freely with her husband's younger brother. A man and a woman in such a relationship [can] sprinkle coloured water on each other at the time of the festival of Holi [essentially a fertility ceremony]" (p. 20); "[in Orissa, in castes unspecified, but presumably 'all poorer classes'] a woman does not speak or show her face to the husband's elder brother, while she can cut the most obscene jokes with the husband's younger brother" (p. 193). This, known to ethnologists as the joking/jesting relationship', is reported as widespread over India, obtaining also between cross-cousins in some south and central Indian communities which 21 practice 'Dravidian' cross-cousin marriage. Since the

21 For wider coverage, see Hutton 1946: 55-56, and Mandel-

baum 1970: 65-66. For linguistic manifestations in some communities (the persons involved in the relationship are 'those

with whom one laughs'), see Emeneau's forthcoming paper "Joking Relationships in India: A Linguistic Note."

EMENEAU

and

VAN NOOTEN:

The Young Wife and Her Husband's Brother

489

with none of the five brothers is that between a wife and her brother-in-law-she is wife to all of them and goes through the marriage ceremony with each. There are in the pertinent passages no occurrences of niyoga phraseology or of the word devara-. We should note that there is parallelism between the ethnological changes discussed above, concerning niyoga and the jesting relationship, and, on the other hand, the historical development of the meaning of devr'-/devara-. Ethnologically, the niyoga institution described in the smrti texts involved all brothers of the widow's husband regardless of their relative age, while in modern times both niyoga and the jesting relationship involve only the husband's younger brother. The Vedic passages discussed are not explicit about which brothers-in-law are intended by devi-. The niyoga-like RV 10.40.2 might well be interpreted as including all brothers-in-law regardless of age, as in the smrti descriptions. On the other hand, if devikamd of AV and RV 10.85.44 is to be interpreted as referring to a jesting relationship with sexual overtones, it seems probable that that of modern times, which is allowable only to the younger brother-in-law, can be the only possible close parallel since the smrti texts allow only sexual relationship "without love" in the niyoga institution. Of course, it might be argued that the dvijdti compilers of the Manu text knew the jesting relationship but found it abhorrent and remained completely silent about anything except their chief concern, which was the provision of issue to the dead man by niyoga, but "without love." As to the semantic development of devara-, it is enough to state here (and give details below) that in the modern vernaculars (modern IndoAryan; NIA) the continuants of devara- mean in most of the languages 'husband's younger brother', usually parallel with the occurrence of niyoga and jesting relationship for this younger brother-in-law only. Evidence for the chronology is as follows. Another fairly early text provides some evidence. It is Hala's Prakrit anthology, the GdthdsaptaSatl, otherwise referred to as the SaptaSatakam or the Sattasai. The chronology of this collection is very uncertain. The editor, Weber (1870, 1881), and other scholars found that identification of the anthologist either with a Satavahana king Hala, who is uncertainly dated about A.D. 68, or with a king Hala of Pratisthana, circa A.D. 467, was most uncertain and in the end improbable. Keith (1920: 40, 224-25) ventured a period A.D. 200 to 450, chiefly on the basis of the nature of the Prakrit. Weber ended with the 3rd century A.D. as the earliest possibility. The only nearly firm date, a terminus ante quem, is provided by a mention of the anthology in the introductory verses of Bdna's Harsacarita; Bdna lived in the

first half of the seventh century A.D. But the Sattasar was a collection which in its various recensions acquired accretions, and (as Keith says, p. 225): "Bana knew [the] collection, and even then we have no security for the existence even in Bana's time of any particular part."22Even this unsatisfactory chronology puts the anthology firmly in a period later than the Vedas, epic, and the smrti texts, but earlier than the commentaries on Manu and on the early dictionaries that we shall have occasion to cite. The anthology is a collection of lyric verses, most of which are erotic in subject and placed in a village milieu. Ten verses are closely concerned with the relation between a young wife and her devara (Prakrit deara-, diara-), and this is the jesting relationship. There are differences of detail in the verses: e.g., (i.28) she shows physical signs (scratches, etc.) of having engaged in love-play with him, (v. 69) she shows jealousy because he has gone to make love with another village girl (the headman's daughter), (i.35) she points to a painting on the house wall of Rama and Laksmana in order (seriously or otherwise) to warn him off making love to her, etc., etc. This is the jesting relationship of modern times, and in fact the earliest in the record apart from the possible Vedic devikdmd. As to the relative age of the devara and her husband, there is in the verses themselves no clue. Weber (1870: 83, n. 1, on vs. i.28) thinks that the relationship itself may point to the devara as the younger brother-in-law, and translates (here and at vs. i.35) devara- as "Schwagerlein"! But in later verses he abandons this for "Schwager." The Bengali translator into English (Basak 1971) usually has "brother-in-law (husband's younger brother)," but there is no warrant for this in the text (as said above), and we may guess that he is translating the Bengali continuant of devara- rather than the Prakrit. We are nevertheless tempted to think that he may be right and that the jesting relationship is already in all details that of modern times-but unfortunately this can be only a guess, even if a plausible one.23 Roughly contemporary with the evidence in the SattasaT are two passages in Pali texts. The Pali Text Society dictionary (Rhys Davids and Stede 1921-25: 330) for devara- gives the meaning "husband's brother, brother-in-law" with a reference to the commentary to the Vimanavatthu, story 32, vs. 6, a canonical verse

22 Hart 1975: 3 follows Keith in stating an approximate date for the SattasaT. 23 In vs. 872 (Weber 1881), not a verse of the original text, the husband is addressed, and the devara is referred to merely as hid tuham 'your brother' (hda = bhda < bhrdtd).

490

Journal of the American Oriental Society 111.3 (1991) The Pali dictionary gives another reference for devara-, viz., "J vi. 152,"which is Fausb0ll's edition of the Jatakas and their commentary, vol. VI (1896), p. 152; the jataka is Khandahdlajdtaka, numbered 542. In the context the princess Canda's husband (the son of GotamT) is to be killed in sacrifice, and this verse (numbered 710) is said to her: ma tvamcanderuccimaranam, bahukatava devara, visalakkhi tamramayissanti te yitthasmim gotamnputte Canda!do not choosedeath!Yourbrothers-in-law are many. 0 large-eyedone! they will cause you delight (or, will gladdenyou) when Gotaml's son has been sacrificed. The commentary has: devard ti patibhdtukd 'devard (means) husband's brothers'. This is explicable in terms of the niyoga and/or the jesting relationship, but adds nothing on the relative age of the devara. Unfortunately, the chronology of the text is even more uncertain than usual; Winternitz (1933: 156) says of the Pali jataka texts: "they cannot serve as documents for the social conditions at the time of Buddha, but at the most, for the period of the 3rd century B.C. [for an uncertain number of the verses], and for the greater part, especially in their prose, only for the fifth or sixth century A.D." There is no possibility of being sure that this verse is any older than the Vimdnavatthu passage. That it is to be referred to North Indian social conditions is clear, as was the Vimdnavatthu passage. We have, then, Pali evidence whose terminus ante quem is about the sixth century A.D. The more firmly fixed chronological data that we have found are the following. The commentary by Ks-irasvamin on the dictionary Amarakosa (Sharma and Sardesai 1941: 140) has on 3.6.32 (bhrdtarah ... svdmino devrdevarau) this gloss: svdminah patyuhbhrata kanisthah 'younger brother of the husband'; KsTrasvdmin'speriod is second half of the eleventh century A.D. As was noted above, Vijfnanesvara's Mitaksara commentary on the lawbook Ydjiavalkyasmrti 1.68 defined the devara who could be appointed for niyoga as kanrydn bhrdtd'younger brother'; this, dated by Derrett (1973) to 1120-1125 A.D., is the earliest datable reference for the modern type of niyoga, even though it is a little later than the K~slrasvamin date for the modern meaning of devara-. Hemacandra in his dictionary Abhidhdnacintamani, dated between A.D. 1160 and 1172, in verse 553 has: devrdevarau/devd va 'varajepatyur 'devr-, devara-, or devan- (in the mean-

which contains this and several other kinship terms denoting persons who are pleased by a young wife's dancing. The commentary on the verse, in Dhammapala's ParamatthadlpanT (Hardy 1901: 135), has the usual derivation of devara- from \/div 'to play' and then: dutiyo varo ti. . . devaro, bhattu kanitthabhata. This, in connecting our word with dutiyo varo 'second husband', obviously parallels the passage quoted above from Yaska's Nirukta. It cannot be derived from that passage since that is taken to be an interpolation later than the thirteenth century A.D. (see above), and Dhammapala is placed (with some dispute; Winternitz 1933: 197, n. 1) either at the close of the fifth century A.D. or sometime in the sixth century. Dhammapala must have drawn his etymology from some earlier source (vedanga or unadi, which we have not found). On the other hand, his gloss of devara- as 'husband's younger brother' (bhattu kanitthabhita) is the earliest datable instance of that meaning that we have found. Since Dhammapala's place of origin is in Tamilnad and the Sinhalese kinship system was at his date already of the Dravidian 24 type, without a special term for husband's younger brother, we must conclude that he derived his gloss from some North Indian source in which this meaning of devara- was already found, a source that, like Ceylon Buddhist sources in general, came from North India. If this is a correct conclusion, it provides us with our earliest date for the semantic development. We had already been tempted to find it as a possibility in the Sattasar, of possibly slightly earlier date. Considering the content of the Vimanavatthu verse (the girl by her dancing delights her husband and her devara-), we should conclude that we are in the same milieu as that of the Sattasai verses involving the devara- and his jesting relationship with his brother's wife. This Pali passage is evidence for the semantic development of devara- older by at least five centuries than that found in the commentaries on the lawbooks and the dictionaries. It, if admitted, and the Sattasar evidence for the jesting relationship put adhesion of the semantic shift and the jesting relationship back to the same period, roughly within or towards the end of the first half of the first millennium A.D.
Trautmann 1981: 153: "The evidence taken as a whole requires that we regard the Sinhalese as a Dravidian community in respect of kinship in the present and as far back as the historical record reaches"; see also pp. 154f. 320-30 (and Emeneau 1939). The Pali data on which he relies are from the fifth century A.D. or earlier. In Turner's 1966 entry for dev&-, there is no Sinhala continuant recorded.
24

EMENEAUand VAN NOOTEN: The Young Wife and Her Husband's Brother

491

ing) husband's younger brother' (Boehtlingk and Rieu 1847: 101). Much later commentaries on the Unddisuitra give similar definitions. E.g., a commentary of unknown author and quite uncertain, though late, date, DaSapddyunddivrtti (Yudhisthira Mimarmsaka 1943: 330, on 8.62), has for devara- the gloss patibhrdtd kanTydn'husband's younger brother'; the editor points out that in Mbh. 1. 106.2 devara- refers to an elder brother and cannot have the specific meaning 'husband's younger brother'. Our modern dictionaries are of little or no help in this matter. PW has for devara- "des Mannes Bruder, insbes. ein jungerer." Similarly, Monier-Williams' dictionary (1899), following P W, has for devi-/devara- "a husband's brother (esp. his younger brother)." This gives no help towards a chronology. Apte (1890) gives for devr- and devara- "a husband's brother (elder or younger)" (with Manu reference). A later, revised edition (of date about 1958; the copy is damaged) gives this for devarah, but for devr "a husband's brother (especially younger)"! The Prakrit dictionary by Sheth (1963), under the entry diara-, gives the meaning (in Hindi) pati kd chotd bhdT 'husband'syounger brother'. The referencesunder devara-, deara-, and diara- yield nothing on the relative age (the SattasaTreference has been examined above), and it is obvious that Sheth has derived his meaning from that of the NIA continuants. Karve (1965: 147, 154) gives in tabular form the NIA material for many of the languages, from Sindhi to Oriya. It is clear that in almost all these languages a distinction is made between the elder and the younger brothers of the husband, and that the latter are referred to by continuants of devara-. The elder brother-in-law is usually designated by some derivative of jyestha'elder' (Turner 1966, entry 5286; for Hindi, Karve gives jeth devar), or of bhrdtrsvasura- (Turner 1966, entry 9669; "brother [who is highly respected like] fatherin-law"; e.g., Hindi bhdsur). In her lists the Bengali forms deor and bhdsur (found in Turner) are not reported. Karve (1965: 190) reports that in "old Marathi literature" there were two terms, but that since niyoga relationships are not found in most of Maharashtra (influence from the Dravidian south), the term dTra means brother-in-law in general. Turner (1966, entry 6546) covers many more modern languages than Karve does, but his meaning entries, though in general coinciding with hers, are much more condensed in format and in general not so satisfactory as hers. Most unexpectedly, he gives "husband's younger brother" for Sanskrit devf-, with RV as the

(earliest) reference; this is most probably by extrapolation from the later Indo-Aryan meanings and is surely not to be accepted.25 Of the chronological questions that we have addressed, we have, we think, found the niyoga practice in RV 10.40.2, and the jesting relationship in AV 14.2.17, 18 and probably in RV 10.85.44. For the semantic development of devi- and its continuants from 'husband's brother' to 'husband's younger brother', the earliest datable evidence is that in Dhammapala's Pali commentary of the fifth-sixth century A.D.; after that there is a gap of about half a millennium before evidence reappears in the Sanskrit commentators on the smrti texts and the dictionaries. We have not directly addressed the problem of the reason for this development. It is undoubtedly to be found in the 'respect' system as it operates between brothers. For the wife certainly, her husband's elder brother is an object of 'respect' and avoidance-he is equated with her father-in-law, as Karve tells us (p. 135, quoted above),26 and as we have seen in the derivation of the Hindi and Bengali term bhasur (Turner 1966 bhrdtrsvasura-). We would assume that the difference in r6le between the brothers vis-a'-vis the wife has been the (perhaps only) factor in producing the semantic difference between terms for 'husband's elder brother' and 'husband's younger brother', relegating the continuants of devara- to the latter meaning and producing new terms for the former with its added 'respect' connotation. Whether we have found all the chronological evidence bearing on this is of course still uncertain.
Two Mahabhdrata passages,27 which have a bearing

on the matters we have been discussing, are particularly
2 In his 1960/1975 article he gives the meaning 'husband's younger brother'; in the last paragraph he says: "Both [i.e., devi- and sydcl-] remain over almost the whole Indo-Aryan domain..., almost exclusively referring to the younger brother in each case." 26 Trautmann 1981: 272f. stresses that "the kinship of the Dharmagastra is shot through with the principle of hierarchy." For his further remarks on the pair of terms under discussion, see p. 98 top. Mandelbaum 1970: 63-66, in a section on the mutual r6les of brothers in the family, deals with the topics which we have raised in this paper and, in particular (pp. 64-65), with the avoidance between a wife and her husband's elder brother and, on the contrary, the joking relationship between her and her husband's younger brother. 27 The anonymous evaluator of the paper has called these passages to our attention.

492

Journal of the American Oriental Society 111.3 (1991) The other relevant passage is MBh. 1.205.27: guror anupravego hi no 'paghato yaviyasah yaviyaso 'nupravegojyesthasya vidhilopakah For entering in to an elder (brother) is no wrong in a younger (brother);entering in to a younger is transgression against the rule in an elder.

difficult to evaluate since they occur in connection with the highly problematical polyandry of the Pandavas. indicates omitted vocatives): MBh. 1.188. 1(... yavIyasah katharn bharyarnjyestho bhrata ... ... samabhivarteta sadvrttah sarns ... How could the elder brother, if he is of good conduct, approach the wife of a younger brother?

The verb sam-abhi-vrt 'to approach' undoubtedly is to be evaluated as a euphemism for 'to have sexual intercourse with', but hardly refers specifically to marriage. This was said by DraupadT'sbrother in an open meeting in which the proposed polyandrous marriage was discussed. Yudhisthira, the eldest of the Pandavas, answered by saying that their mother's statement that all five brothers should in common enjoy the alms they had brought home (Draupadl!) must not be falsified. This says nothing about the highly anomalous decision that each brother should separately and successively marrythe bride. This verse, however, can surely be considered a statement indicative of custom in the epic period, and it, with its unstated converse (that a younger brother could "approach"the wife of an elder brother), seems quite parallel with the modern custom that we have discussed above and with the niyoga of the earlier period as it was interpreted later. We have then an additional piece of chronological evidence for the distinction between elder and younger brothers and their access to one another's wives, earlier than any of the evidence reviewed above, though in fact we cannot be sure of the relative age of evidence from our critical epic text and that from the Pali sources.

This verse, with its condensed style, is only interpretable in terms of the context, which contains the agreement arranged by the Pandava brothers (1.204.28) to prevent quarrels, viz., that, if one of the brothers should see another sitting (samdsina-) with Draupadl, he (the one who sees) should become a brahmacarin in the forest for twelve months.28 Arjuna sees Yudhisthira with Draupadi, and Yudhisthira, with a number of verses including this quoted one, tries to dissuade him from following the agreement. It is difficult to interpret the verse in terms of anything that we have found so far. We are tempted to think of an implication of the verse, viz., that it is wrong for a wife to be in the presence of an elder brother of her husband. Above we have seen that in modern North India this is indeed disrespectful and wrong on the wife's part. If this interpretation is correct, this is very early evidence for this item of social structure, which we have adduced as the reason for the shift in the meaning of the continuants of devi-.
28 Reading masani (or mdsdn plus a syllable preceding or following) with the southern Mss instead of the (rather nonsensically improbable) varsdni of the critical edition with its wavy line; see van Buitenen's note to his translation.

REFERENCES

Apte, Vaman Shivram. 1890. The Student's Sanskrit-English Dictionary.... Poona: Shiralkar and Co. Aufrecht, Theodor. 1861, 1863. Die Hymnen des Rigveda. Indische Studien, 6-7. Berlin: Ferd. Dummler. Barret, LeRoy Carr. 1938. The Kashmirian Atharva Veda, book eighteen, edited with critical notes. JAOS 58: 571614. Basak, Radhagovinda. 1971. The Prdkrit Gatha-saptasat!... edited with ... translation.... Bibliotheca Indica, 295. Calcutta: The Asiatic Society. Bloomfield, Maurice, and Franklin Edgerton. 1932. Vedic Variants ... vol. II: Phonetics. Philadelphia: Linguistic Society of America.

Bloomfield, Maurice, and Richard Garbe. 1901. The Kashmirian Atharvaveda (School of the Paippaladas) reproduced by chromophotography from the manuscript.... 3 vols. Baltimore: Johns Hopkins Press. Bohtlingk, Otto, and Charles Rieu. 1847. Hemakandra'sAbhidhdnakintdmani... herausgegeben,ubersetzt.... St. Petersburg: Kaiserliche Akademie der Wissenschaften. Buhler, Georg. 1886. The Laws of Manu, translated.... Sacred Books of the East, 25. Oxford: Clarendon Press. Burrow, Thomas. 1955. The Sanskrit Language. London: Faber and Faber. 1973. Sanskrit pd "go, move, pass, traverse." IIJ 15: ._ 81-108.

EMENEAUand VAN NOOTEN:The Young Wife and Her Husband's Brother Delbrick, Berthold. 1889. Die indogermanischen Verwandtschaftnamen, ein Beitrag zur vergleichenden Alterthumskunde. Abhandlungen der philologisch-historischen Classe der konigl. Sachsischen Gesellschaft der Wissenschaften XI.5. Pp. 1-228/377-706. Derrett, J. Duncan M. 1973. Dharmagdstra and Juridical Literature. A History of Indian Literature (ed. Jan Gonda), part of vol. IV. Wiesbaden: Otto Harrassowitz. . 1978. The Death of a Marriage Law: Epitaphfor the Rishis. New Delhi: Vikas Publishing House. Emeneau, M. B. 1939. Was there Cross-cousin Marriage among the 9akyas? JA OS 59: 220-26. . (Forthcoming.) Joking Relationships in India: a Linguistic Note. Fausb0ll, V. 1896. The Jdtaka together with its Commen... , vol. VI. London: Kegan Paul Trench Trubner tary and Co. Geldner, Karl Friedrich. 1951. Der Rig- Veda. . . ins Deutsche ubersetzt.... 4 vols. Harvard Oriental Series, 33-36. Cambridge, Mass.: Harvard University Press. Grassmann, Hermann. 1873. Worterbuchzum Rig-veda. Leipzig: F. A. Brockhaus. . 1876-77. Rig-veda, uebersetzt.... 2 vols. Leipzig: F. A. Brockhaus. Haas, E., and Albrecht Weber. 1862. Vedische Hochzeitsspruche. Indische Studien 5: 177-266. Hardy, E. 1901. Dhammapala's Paramattha-dipani, part IV: Being the Commentary on the Vimdna-vatthu. Pali Text Society [vol. 48]. London: Pali Text Society. Hart, George L., III. 1975. The Poems of Ancient Tamil. Berkeley and Los Angeles: University of California Press. Hutton, J. H. 1946. Caste in India: Its Nature, Function, and Origins. Cambridge: Cambridge University Press. Jolly, Julius. 1885. The Institutes of Narada .... Bibliotheca Indica, 102. Calcutta: Asiatic Society of Bengal. . 1887. Mdnava Dharma-gdstra,the Code of Manu ... text critically edited. London: Trubner and Co. . 1889. The Minor Law-books, part I: Ndrada.... Sacred Books of the East, 33. Oxford: Clarendon Press. Kane, Pandurang Vaman. 1941, 1946. History of Dharmasastra, vols. 2, 3. Government Oriental Series, class B, no. 6. Poona: Bhandarkar Oriental Research Institute. Kangle, R. P. 1963. The Kautillya Arthasastra, part II: An English Translation. Bombay: University of Bombay. Karve, Irawati. 1965. Kingship Organisation in India. 2nd ed. Deccan College Monograph Series, 11. Poona: Deccan College Post-graduate and Research Institute. Keith, A. Berriedale. 1920. A History of Sanskrit Literature. London: Oxford University Press. Lakshman Sarup. 1921. The Nighantu and the Nirukta, vol. 2: English Translation and Notes. Oxford: Oxford University Press.

493

1927. The Nighantu and the Nirukta, vol. 1: Sanskrit Text. Lahore: University of the Punjab. Langlois, M. 1848-51. Rig-Veda. .. traduit du sanscrit. 4 vols. Paris: Librairie de Firmin Didot Freres. Ludwig, Alfred. 1876-88. Der Rigveda ... ins Deutsche iubersetzt mit Commentar und Einleitung. 6 vols. Prag: F. Tempsky. . 1889. Ober die Kritik des 1RgvedaTextes. Abhandlungen der k. bohm. Gesellschaft der Wissenschaften, VII Folge, 3 Band, philos.-histor. Cl. 5. Macdonell, Arthur Anthony. 1904. The Brhad-devata Attributed to ?aunaka. 2 vols. Harvard Oriental Series, 5-6. Cambridge, Mass.: Harvard University Press. Mahdbhdrata. 1933. The Mahdbhdrata... critically edited, by Vishnu S. Sukthankar..., vol. 1 (Adiparvan). Poona: Bhandarkar Oriental Research Institute. . 1961. The Mahdbhdrata. . . critically edited, by Vishnu S. Sukthankart and S. K. Belvalkar . . . , vol. 13 (Sdntiparvan, 1st part, ed. S. K. Belvalkar). Poona: Bhandarkar Oriental Research Institute. . 1966. The Mahabhdrata ... critically edited, by Vishnu S. Sukthankart, S. K. Belvalkar, and P. L. Vaidya. .. , vol. 17 (Anusasanaparvan, ed. R. N. Dandekar). Poona: Bhandarkar Oriental Research Institute. Mandelbaum, David G. 1970. Society in India. 2 vols. Berkeley and Los Angeles: University of California Press. Monier-Williams, Monier. 1899. A Sanskrit-English Dictionary.... 2nd ed. Oxford: Clarendon Press. Muller, F. Max. 1849-74. Rig-veda-sanhita. . . with the Commentary of Sayana. 6 vols. London: Wm. H. Allen and Co. . 1873. The Hymns of the Rig-veda ... reprintedfrom the editio princeps. London: Trubner and Co. Oldenberg, Hermann. 1886. The Grihya-Satras, Rules of Vedic Domestic Ceremonies, translated, part 1: ... Asvalfyana-g~jhya-satra.... Sacred Books of the East, 29. Oxford: Clarendon Press. . 1888. Die Hymnen des Rigveda.. ., Band I: Metrische und textgeschichtliche Prolegomena. Berlin: Wilhelm Hertz. _ 1912. Rgveda. Textkritische und exegetische Noten: Siebentes bis zehntes Buch. Abhandlungen der k. Gesellschaft der Wissenschaften zu Gottingen, phil.-hist. Klasse, n.F., Bd. XIII, 3. Prabhati Mukherjee. 1963. Widowhood and Niyoga in the Arthasastra and Manusmrti. Our Heritage 11.1: 2nd paper, 11 pages. PW = Bohtlingk, Otto, and Rudolph Roth. 1859-61. SanskritWorterbuch. . . , 3ter Teil:jia-dha. St. Petersburg:Buchdruckerei der K. Ak. d. W. Ray, Pratap Chandra. 1883-96. The Mahdbhdrata ... translated into English prose.... Calcutta: Bharata Press.

494

Journal of the American Oriental Society 111.3 (1991) veshvaranand Indological Series, 15. Hoshiarpur: Vishveshvaranand Vedic Research Institute. Weber, Albrecht. 1870. Ueber das Saptaqatakam des Hala. Abhandlungen fur die Kunde des Morgenlandes, V.3. Leipzig: F. A. Brockhaus. __ . 1881. Das Saptaqatakam des H6la. Abhandlungen fur die Kunde des Morgenlandes, VII.4. Leipzig: F. A. Brockhaus. Whitney, William Dwight, and Charles Rockwell Lanman. 1905. Atharva- VedaSarhhitd,translated.... 2 vols. Harvard Oriental Series, 7-8. Cambridge, Mass.: Harvard University Press. Wilson, H. H. 1850-88. Rig-veda Sanhita ... translated. 6 vols. London: Trubner and Co., Wm. H. Allen and Co. Windisch, Ernst. 1920. Geschichte der Sanskrit-Philologie und indischen Altertumskunde, 2terTeil. Grundriss der Indo-Arischen Philologie und Altertumskunde, l.l.B. Berlin-Leipzig: Walter de Gruyter. Winternitz, Maurice. 1897. Notes on the Mahabharata with special reference to Dahlmann's "Mahabharata."JRAS 1897: 713-59. . 1933. A History of Indian Literature, vol. II, translated by Mrs. S. Ketkar and Miss H. Kohn. Calcutta: University of Calcutta. Yudhishthira Mimansaka. 1943. Daqapddyunddivrtti.Princess of Wales Sarasvati Bhavana Texts Series, 81. Benares: Indian Press.

Rhys Davids, T. W., and William Stede. 1921-25. PaliEnglish Dictionary. London: Pali Text Society. Roth, Rudolph, and W. D. Whitney. 1855-56. Atharva Veda Sanhitf, herausgegeben. Berlin: Ferd. Dilmmler. . 1924. AtharvaVedaSanhita.2te verbesserte Auflage besorgt von Dr. Max Lindenau. Berlin: Ferd. Dummler. Sharma, Har Dutt, and N. G. Sardesai. 1941. Amara's Ndmalinganusasanam with the Commentary AmarakoSodghatana of Bhatta Ksirasvdmin. Poona Oriental Series, 43. Poona: Oriental Book Agency. Sheth, Hargovind Das T. 1963. Paia-sadda-mahannavo: A Comprehensive Prakrit-Hindi Dictionary.... 2nd ed. Prakrit Text Society Series, 7. Varanasi: Prakrit Text Society. Szemerenyi, 0. 1977. Studies in the Kinship Terminology of the Indo-European Languages, with special reference to Indian, Iranian, Greek and Latin. Acta Iranica 16 (3me ser., vol. VII): 1-240. Trautmann, Thomas R. 1981. Dravidian Kinship. Cambridge Studies in Social Anthropology, 36. Cambridge: Cambridge University Press. Turner, R. L. 1960. Sanskrit gvasura-. Bulletin of the School of Oriental and African Studies 23: 106-8 (= 1975. Collected Papers 1912-1973, pp. 390-93. London: Oxford University Press). . 1966. A Comparative Dictionary of the Indo-Aryan Languages. London: Oxford University Press. Vishva Bandhu. 1961. Atharvaveda (Saunaka) with the Padapart III. Vishpdtha and Saya~na'sCommentary...,