Strīkāmā vai gandharvāḥ. Spirit-possession, women, and initiation in Vedic India

The nature of the Vedic gandharvas, and their female counterparts, the apsarases, has been the subject of much controversy. While often appearing in Vedic texts as a spirit of procreation, opinions have differed as to whether the pali gandhabba-, mentioned in the Buddhist Nikāyas as a being whose presence is necessary for the conception of a human being, should be understood in this light, or (following Buddhist commentarial tradition) as a disembodied soul entering the womb in order to reincarnate. The present study, opting for the first alternative, will explore the older, Vedic gandharva’s nature not only as a genius of fertility and procreation, responsible for conception as well as miscarriage, but also as a lusty, potentially harmful spirit with a taste for mortal women – even married ones. The gandharva’s (in post-RV texts usually in plural) desire for women sometimes expresses itself in his taking possession of mortal females; thus explaining text-passages where gandharvas are said to cause mental illness. This possession could however have positive aspects, as the gandharva could speak through the possessed woman and leave oracular answers on questions of esoteric and ritualistic matters. This belief, it will be argued, shows affinities to possession cults around the world, where the role as possessed oracle is frequently played by girls or women – a fact well-known from anthropology. But gandharva-possession must also be understood in the light of a conception of the gandharva as mediator between this world and the other; as a being with a knowledge of divine secrets, which he could impart to mortals in exchange for the enjoyment of their women. This same kind of exchange often appears in myths, such as that of the contest between gods and gandharvas for the soma and the goddess Speech. In mythology, the gandharvas and apsarases are frequently depicted as youthful, goodlooking, and fond of games and sports, music and dancing, and erotic activities. A case will be made for a connection between these beings and the adolescent period of life in Vedic society. In the domestic ritual, offerings are made to “the gandharva” by girls about to get married, thus asking for his permission; while young men who have completed their studies and are about to settle down and marry, are given a staff representing the gandharva, to “guard” them in their new life. All this suggests that the single gandharva of older Vedic times was thought of as a tutelary deity of unmarried adolescents, male as well as female (the latter being “married” to him). There are also instances in the priestly ritual where “young, beautiful” boys and girls impersonate the gandharvas and apsarases. Some sort of bond between these deities and people in the adolescent, pre-married stage of Vedic life (the prathama-vayasáor “first age-span”), is also suggested by the association of these beings 1 I would like to thank professor emeritus Folke Josephson for reading a draft of this article, and for checking the references to ancient Iranian texts.

repeated in texts like Vasubandhu's Abhidharmakośabhāṣya, while the Amarakośa and other classical Sanskrit dictionaries gloss the word with the distinctly Buddhist term antarābhavasattva-, a "being of the intermediate state", specifying that this means a being in the state between two incarnations, i.e., a deceased individual waiting to be born in a new reincarnation. 7 This is, indeed, the meaning of the antarābhava-or "intermediate state" of learned Buddhist speculation. 8 How are we to evaluate this younger tradition? Scholarly opinion on the matter has mainly been divided along two lines, between those who accept the traditional interpretation, and those who see no support for this in the older Pali texts, preferring instead to see the gandhabba here as a genius of procreation. Among the latter we find Lord Chalmers, who, without taking notice of the traditional interpretation, rendered the word as "presiding deity of generation" in his translation of MN; 9 as well as the Pali Text Society's Pali-English Dictionary in its entry on gandhabba, which, referring only to MN and the Milindapañha, states that this being is "said to preside over child-conception". 10 The other line of interpretation seems to be by far the most widespread one. First, as it seems, put forward by Oldenberg 11 and Windisch 12 , the view of the gandhabba as a disembodied being in between two incarnations has been accepted by a series of nameworthy scholars. 13 It is thus argued that, while the MN does not explicitly describe the being as a deceased individual awaiting its next incarnation, the words paccuppaṭṭhito hoti ("is present") suggest that it is not merely a deity presiding over procreation, but is actually taking part in the act in some way; and further, Asita Devala's question to the brahmins, whether they know what class the gandhabba might have belonged to (which they don't), followed by the question, "In that case, sirs, do you know who you are?", is seen as clear evidence that the gandhabba is the very being that is reborn (the brahmins not knowing what class they belonged to in their former lives). While acknowledging that early Buddhism had no concept of a transmigrating soul, a few scholars have sought to identify the gandhabbha with the consciousness, viññāṇa-, which is elsewhere in the Nikāyas said to survive the destruction of the body and "descend" (ava-kam-) in a womb in order to be reborn. 14 Assuming that this interpretation is correct, one important question remains: what connection, if any, is there between this Buddhist conception and the older, Vedic concept of the gandharva -considering that the belief in transmigration does not appear in Vedic literature until the early Upaniṣads, and is not even there in any way connected with the gandharvas? Those scholars who have addressed this issue have mostly pointed out the Vedic gandharva's function as a fertility deity, with a somewhat unclear connection to marriage and conception, and have seen the Buddhist notion as a development of these ideas, having been reinterpreted under influence of the transmigration doctrine. Keith, for instance, points to a in brahmanical as well as Buddhist mythology, it is nowhere else said to be a discarnate spirit awaiting its next incarnation. Remarkably, the numerous discourses on transmigration and karman in the Pali and later Buddhist texts do not include the gandhabba in the role of a being in a state between two incarnations. 23 Hillebrandt summarizes: Es ist also gegenüber der gesamten Literatur und den Kunstzeugnissen eine verschwindende Anzahl von Stellen, die für diese absonderliche Bedeutung von gandharva eintreten. Die stellen im Majjh. Nikāya, Assalāyanasutta und Divyāvadāna sind sehr ähnlich und scheinen ein freies Zitat zu sein, das aus einem älteren Text entlehnt ist. Im Assalāyanasutta soll bewiesen werden, daß zwischen den verschiedenen Kasten kein wesentlicher Unterschied bestehe … Es ist nun keineswegs ausgemacht, daß ,der Faden der Argumentation, wie Oldenberg meint, 24 total zerschnitten würde', wenn wir Gandharva hier als einen Genius der Fruchtbarkeit ansehen wollten, der ja ebenfalls keiner Kaste angehören würde … 25 Rejecting, thus, the view of the gandhabba in MN as a spirit entering the womb, Hillebrandt identifies the Milindapañha passage as the earliest text expressing this belief; it must, he suggests, have been composed when the MN passages (which it quotes) were no longer fully understood, and the ancient conception of the gandharva as a fertility deity had become obsolete. "Mir scheint daher, daß die Auffassung der Milindapañho nichts anderes ist als ein Mißverständnis brahmanischer Mythologie." 26 I agree with Hillebrandt's conclusions. A close look at the contexts of the two MN passages raises serious objections to the traditional interpretation of these. We will first discuss the one in the Mahātaṇhāsaṅkhayasutta, which has not received as much comment as the passage in the Assalāyanasutta. What is the context that makes the Buddha bring up the topic of procreation, including the gandhabba, in this sutta? The reason for his entire sermon here is the heretical view propagated by a certain monk, who holds that a transmigrating being is constituted by one and the same consciousness (viññāṇaṃ), which enters ever new bodies. This view being completely incompatible with the Buddhist "no-self" doctrine, which denies any lasting soul or essence that transmigrates, the Buddha forcefully rejects the monk's belief, demonstrating how an individual is actually a conglomerate of numerous constituents, put together by various causes. The sermon mainly consists of an exposition of the "conditioned arising" (paṭiccasamuppāda-), the chain of causes responsible for the rebirth and formation of an individual. At the end, this teaching is illustrated with an exposition on how these causes work in a person's life, beginning with conception; and it is here that we find the passage on the gandhabba. The entire sermon is, thus, intended to reject the theory of a transmigrating being; each individual creature is the result of numerous causes -among which we find "the three things" facilitating conception: the parents, and the gandhabba.
The case is similar with the Assalāyanasutta. While, as we have seen, Asita Devala's question as to the social class of the gandhabba has been seen as confirming the theory of an incarnating spirit, Devala does, in fact, question the brahmins in a similar vein on the social class of their mothers, fathers, and ancestors: do they know if their mothers have been only with their (brahmin) husbands, or their fathers only with brahmin women? Or if their mothers' mothers, and fathers' fathers, seven generations back, have only been with men or women belonging to the same class? And, finally, do they know the class of the gandhabba, the third cause (after the parents) of their conception? The entire discourse is meant to humble the haughty brahmins, who consider their own class the very highest.
Finally, some words should be said on the "hopelessly confused" 27 presentation in the Milindapañha. 28 While even Hillebrandt acknowledged that this text connects the gandhabba with rebirth, its account is remarkably dissimilar from the later ones, which identify the gandhabba as a being of the "intermediate state". Instead, we appear to be facing two different definitions of the word in this text, as it attempts to clarify the old MN passage. We are given a description of how "a certain gandhabba, coming from somewhere or else", 29 enters a womb (yoni-) in a family of some species (egg-born, "sweat-born" etc.). The being is not specified to be a spirit between two incarnations, and the phrasing suggests an uncertainty on the part of the author as to the gandhabba's sphere of origin. This is markedly different from the later account of Buddhaghosa, who has the gandhabba take birth "driven on by the mechanism of karman"; in fact, the Milindapañha does not expressly mention re-birth at all in this context. The cause of the gandhabba's taking birth in some species or another is said to be kulavasena, "by the power of family [or 'species']", perhaps implying that the gandhabba is attracted to various kinds of species; 30 in any case, this is expressly different from birth caused by karman -kammavasena -which is only one of four ways in which the "descent of the embryo" might take place. The birth caused by kulavasa-is the only one of these explicitly involving the gandhabba. 31 All this suggests that the integration of the gandhabba into the transmigration doctrine took place successively, long after the composition of the Nikāyas. This holds especially for its being identified as a being of the "intermediate state" (antarābhava-), a concept which, though fairly ancient, first appears as a topic of the doctrinal disputes of postcanonical Theravāda Buddhism. 32 The Vedic gandharva as a procreation-divinity Now, there are clear indications of a connection between gandharvas and fertility in older, Vedic literature; not least in the spells and incantations of the Atharvaveda. In AV 4.4 (Śaunaka recension), the gandharvas are mentioned as having restored the virility of the god Varuṇa by means of a "penis-erecting herb" (óṣadhiśepahárṣaṇī-, v. 1). An association with 27 Wijesekera 1994a, p. 196. 28 Trenckner's ed., pp. 123ff, 128ff. 29 yo koci gandhabbo yato kutoci āgantvā. 30 "If, thus, the gandhabba, coming from somewhere or else, appears in an eggborn family, he becomes eggborn. If, thus, the gandhabba, coming from somewhere or else, appears in a placenta-born family, he becomes placenta-born." (The same applies to "sweat-born" etc.) (yadi tattha gandhabbo yato kutoci āgantvā aṇḍaje kule uppajjati, so tattha aṇḍajo hoti. yadi tattha jalābuje yato kutoci āgantvā jalābuje kule uppajjati, so tattha jalābuje hoti.) 31 In the case of birth caused by prayer (āyācanavasena; p. 129), a deity (devaputta-) is sent down from heaven by Sakka to take birth in a virtuous family "with many offspring but no sons"; the "deity" does not, however, seem to be identical with the gandhabba. 32 Only after finishing the above discussion did I come upon the important work of Langer (2000), whose first chapter deals with the problem of the gandhabba. Langer's conclusions come close to ours: the gandhabba is deemed to be an ancient spirit of procreation, as evinced by Vedic texts, and the stereotyped formula in the Pali texts is, suggests the author, drawn from popular beliefs or sayings. In later times, it was no longer understood and had to be reinterpreted. virility seems again implied in 4.34.2-4; verses with the purpose of making sure that deceased men will not dwell in the next world bereft of their virile force, but will be able to indulge fully in the celestial pleasures awaiting them. "Jātavedas [the cremation fire] burns not their male organ; in the heavenly world, lots of women-folk are theirs", 33 says v. 2, and the second half of the next verse states about the dead person that he "stays with Yama [the god of the dead], goes to the gods, revels with the soma-drinking gandharvas" (transl. Whitney, slightly modified). 34 The notion of the gandharvas as "reveling" (mad-), often together with the apsarases, recurs several times in AV, as we shall see, and probably has some erotic implications; in light of v. 2, it would seem that the dead man partakes of the same kind of pleasures, and this seems to be confirmed by the next verse (v. 4b), which carries on the theme of preserved virility in stating of the dead men that "Yama robs them not of their seed (retas)". 35 Though not actually bestowing virility in these verses, the gandharvas clearly appear as associated with it; indeed, as these beings do not usually appear in the heaven of the dead, their mention here seems to be facilitated exclusively by their typical indulgence in sensual -especially erotic -pleasures. The same association with virility no doubt underlies the offering, at the aśvamedha-sacrifice, of the penis of the sacrificial horse to the gandharvas (and the testicles to the apsarases) as prescribed in TS 5.7.15; as well as the term mūrdhanvām ̐ s … gandharvaḥ, "the gandharva with the head", used in a wedding hymn as, apparently, a designation for the penis. 36 A fertility function also seems implied in the offering of three measures (mā́trā) of grain to the gandharvas at harvest in AV 3.24.6; a connection with vegetation is echoed in some younger texts. 37 In some Middle Vedic texts, we find the gandharvas and, especially, the apsarases expressly associated with sexual intercourse (mithuna-): ŚB 8.6.1. 21,9.4.1.4;JUB 3.25.8.
The connection with procreation and birth, whatever be its exact nature, is thus a recurring theme in the Vedic depiction of the gandharva. The designation, in a wedding hymn, of the penis as "the gandharva with the head", and, at the rite for conception, of the wife's genitals as the mouth of Viśvāvasu (implying an identification of the gandharva with the womb, or perhaps locating him inside it?), are clearly late echoes of the same kind of conception as encountered in some AV hymns. It may be noted that the notion of a deity responsible for placing and protecting the embryo inside the womb (if something like that is indeed the 33 náiṣāṃ śiśnáṃ prá dahati jātávedāḥ svargé loké bahú stráiṇam eṣām. 34 ā́ste yamá úpa yāti devā́nt sáṃ gandharváir madate somyébhiḥ. 35 Transl. Whitney (in Joshi 2004), slightly modified. (náinān yamáḥ pári muṣṇāti rétaḥ.) 36 The hymn is found in the Kāṭhaka Gṛhyasūtra 25.23; parts of it are quoted in other sūtras (Caland 1929). Mūrdhanvant is also found as a name in the list of gandharvas in Taittirīya Āraṇyaka 1.9. 3 (cf. ibid., p. 309). Of interest in this context is also the derivation of Kandarpa, a name of the god of love in later mythology, from gandharva-, proposed by Barnett (1928, p. 704 n. 2), who states: "I would explain Kandarpa as a Prakrit form of Gandharva. In some of the vernaculars classed together by the grammarian as 'Paiśācī', particularly the Drāviḍī, the word gandharva might either become directly kandarpa, or first change to kandappa and thence by a false etymology from darpa be sanskritised into kandarpa." 37 In Vādhūla Śrautasūtra 4.116, the gandharvas and apsarases are connected with draught-oxen and ploughs, respectively, because "these gandharvas and apsarases generate food (i.e., crops)" (ete gandharvāpsarasa evānnaṃ janayanty). (Text in Caland 1928, p. 237.) ŚB 11.2.3.9 mentions the gandharvas Yavamān ("possessing barley"), Uddālavān ("possessing uddāla-grain") and Antarvān ("pregnant"), as connected with winnowing baskets, agriculture, and grain, respectively. Cf. possibly JB 3.226, where the "Svāśira apsarases" are identified with the herbs (oṣadhayaḥ). function of the gandharva/gandhabba) is not foreign to later Indian mythology, where we find Naigameṣa or Nejameṣa filling this function. 38 The Vedic gandharvas and apsarases have an ambivalent relationship to marriage and wedding-ceremonies. We have already mentioned the staff representing Viśvāvasu, which is placed between the newly-wedded couple during the three nights of abstention. As Slaje has shown, 39 this custom -including the three nights of celibate -is not to be found in early Vedic literature, appearing first in the Gṛhyasūtras. 40 It does, however, have older precedents. Middle Vedic texts tell us that the wife was in a state of impurity not only during her period, but also for the first three days following it; no intercourse was to take place during this period, but only on the fourth night, after the wife had changed her soiled garment. The first time of cohabitation thus occurred as soon as the wife was clean from menstrual impurity; not after a specific number of nights following the wedding. 41 A similar practice is, furthermore, prescribed for the newly initiated Veda-student, who had to spend the first three days after initiation in penance, eating non-salted food (so also the newly-wedded couple) before being fit to learn the Sāvitrī mantra. 42 The concept of three "dangerous" nights, with suspension of certain activity, was thus an integrated part of certain Vedic rites of passage. There is no mystery about how this practice came to be included also in the rites of marriage and procreation; what has to be answered is rather how Viśvāvasu came to be a fundamental ingredient of the rite -neither he nor any other gandharvas having any part in the older practice. Presumably the notion of Viśvāvasu and the use of the staff were added fairly late to the practice, as some Gṛhyasūtras do not include either in the prescription of three nights' sexual abstention; 43 where the staff appears but Viśvāvasu is not mentioned, 44 the prescription that the staff be smeared with perfume (gandhalipta-) has, however, been interpreted as a reference to the gandharva; 45 the frequent association of these beings with fragrance (gandha-) being due to the traditional etymology of the word gandharva-.
The most common, and likely, interpretation of the practice is that the wife during these three nights "belonged to" Viśvāvasu. 46 While the rite is late, the notion of Viśvāvasu claiming his rights to the newly-wedded woman appears very clearly already in RV and AV.
38 So already in RVKh. 4.13.1, and some Gṛhyasūtras; see Winternitz 1895. 39 Slaje 1997 For the practice -the so-called caturthīkarman -cf. in general Kane 1974, pp. 202ff;Pandey 2002, pp. 222ff;Gonda 1980, p. 394. 41 The Gobhila Gṛhyasūtra (2.5.8) still prescribes the end of menstrual impurity as the time for cohabitation, besides this mentioning the three nights of abstention as an alternative being prescribed by "some" (7: eke). 42 Slaje thus rejects the theory of an Indo-European origin of the marriage practice; for the use, in legends and in actual practice, of a stock or a sword as a symbolic sexual barrier between a sleeping couple, cf. e.g. West 2007, pp. 436-37. 43 E.g. Pāraskara Gṛhyasūtra 1.8.21; Jaimini 1.22; Śāṅkhāyana 1.17.5ff. 44 Āpastamba Gṛhyasūtra 8.9; BGS 1.5.16ff. 45 E.g., Oberlies 2005, p. 103. Oberlies (ibid., n. 26) expresses doubt as to the connection with this staff and the one given to the snātaka or Veda student who has completed his studies and is about to return home -this staff, too, being addressed as "Viśvāvasu"; the identification is, according to Oberlies, "nicht unproblematisch", as the snātaka's staff is said to be vaiṇava-, made of bamboo, while the one used during the nights of chastity is to be made of udumbara. While it may be true that this is not unproblematic, the identification of both staffs with Viśvāvasu can hardly be considered a coincidence. It is possible that the snātaka's staff, being mentioned in Gṛhyasūtras which do not yet know the marriage custom (Jaimini 1.19; Gobhila 3.4.27), belongs to the older practice, and has been borrowed into the marriage rites; but what its original meaning was, is far from certain. 46 See for example Keith 2007, pp. 375-76. In the "wedding hymn" RV 10. 85, are directed to Viśvāvasu, imploring him to go away from the bride: Rise up from herefor this one has a husband! I worship Viśvāvasu with obeisance and words of praise. Seek another one, dwelling in her father's house, [though] mature; that is your share by birth -seek it out! Rise up from here, Viśvāvasu! We worship you with obeisance. Find another, attractive one; let the wife unite with her husband! 47 More light on the nature of the relationship between Viśvāvasu and the bride is thrown by the verses 40-41: Soma knew you first, the gandharva knew you next; Agni was your third husband; the fourth one is of human birth. Soma gave you to the gandharva, the gandharva gave you to Agni; Agni has given me wealth and sons, and now this [wife]. 48 Agni, the fire god, is here clearly the nuptial fire; 49 the gandharva thus possessed the bride until the performance of the nuptial ceremonies. Viśvāvasu is now implored to seek another woman, who is still living with her father (pitr̥ ṣádaṃ 50 ). The same conception of Viśvāvasu is found in the Atharvaveda; AV 14.2.33ff implores him, partly with words echoing those of RV 10.85, to go away from the bride and seek out a girl living with her father (v. 33), or return to the apsarases -"those are your kin" (tā́s te janítram, v. 34); "go away toward your wives, the apsarases" ('bhí jāyā́ apsarásaḥ párehi, v. 35). The gandharva, whose lustful nature we are going to explore, has had legitimate claims to the woman before her marriage, but may not give up those claims even after the ceremonies, and has to be made to leave with implorations and spells. 51 The notion of the gandharva's right to unmarried girls may underlie a rite prescribed in the KGS, according to which a girl reaching puberty, and thus becoming marriageable, is to perform a worship that in some way involves gandharvas (and a feminine Gandharvāṇī); 52 the gandharva is also included among the deities to which a young girl sacrifices before leaving her parents' house to get wedded (Śāṅkhāyana Gṛhyasūtra 1.11.4); similarly at KGS 17.1, where "Viśvāvasu the king of gandharvas (gandharvarāja-)" appears at the end of a list of mostly abstract deities (e.g. Kāma, Bhaga, Hrī, Śrī, Lakṣmī, Puṣṭi), to 47 úd īrṣvā́taḥ pátivatī hy èṣā́ viśvā́vasuṃ námasā gīrbhír īḷe / anyā́m icha pitr̥ ṣádaṃ vyàktāṃ sá te bhāgó janúṣā tásya viddhi // úd īrṣvā́to viśvāvaso námaseḷā mahe tvā / anyā́m icha prapharvyàṃ sáṃ jāyā́m pátyā sr̥ ja //. 48 sómaḥ prathamó vivide gandharvó vivida úttaraḥ / tr̥ tīýo agníṣ ṭe pátis turīýas te manuṣyajā́ḥ // sómo dadad gandharvā́ya gandharvó dadad agnáye / rayíṃ ca putrā́ṃś cādād agnír máhyam átho imā́m //. 49 E.g., Oberlies 2005, pp. 101-102. 50 The meaning of this word being uncontroversial, I see no reason for seeing, with Vasilkov (1990, p. 395 and n. 6), the first member pitr̥ -as meaning the Ancestors, and the compound as referring to a Männerbund whose members, called gandharvas, represent or impersonate the spirits of the dead. 51 This is still the case in Bṛhadāraṇyaka Upaniṣad 6.4.19, a text which is in all likelihood older than any of the Gṛhyasūtras; here, in the context of the rites for begetting offspring, Viśvāvasu is asked to go away with (a variant of) the verse RV 10.85.22, no reference being made to any preceding nights of chastity. 52 KGS 19-20. It is not very clear what is meant when the girl is admonished to lit two fires gandharve devakule vā, "in/at a gandharva or a shrine" (19.3). According to Devapāla's commentary, the gandharva-here is a place where water flows without cause (nimitta-); if so, the connection with gandharvas might be due to their association with water (for which see below). "Gandharvāṇī" in 20.2 is invoked with two other deities whose name are but feminized versions of those of male gods -Indrāṇī and Varuṇāṇī -and may conceivably be based on the male gandharva of an older version of the rite. whom offerings are made at the giving away of a girl for marriage. It appears that the girl, upon entering adulthood and a married life, had to propitiate the gandharva, to whom she had previously belonged. When Viśvāvasu or the gandharvas and apsarases in general are worshipped at wedding rites, 53 we may be dealing either with the same kind of belief, or simply with the old association with fertility.
AV 4.37 is a spell for warding off a menacing gandharva, who appears to be approaching a (married?) woman. The general tone of the spell is considerably more violent than the pleadings of RV 10.85 and AV 14.2; the gandharva appears as a more explicitly demoniacal being, 54 and the words used to chase him off are forceful, even describing the physical injuries their magic inflicts upon the lustful demon. Though the spell has been considered yet another series of protective formulas used against the wife-coveting Viśvāvasu, I think this is doubtful; the gandharva is nowhere called Viśvāvasu, nor is there any reference to a recent wedding. The underlying conception may simply be the gandharvas' lust for women, wedded or not. 55 Parts of the spell will be quoted: May the apsarases go to the river, to the ford of the waters […]. Gulgulū, Pīlā, Naladī, Aukṣagandhi, Pramandanī -go away thither, apsarases; you have been recognized! Where there are fig-trees, banyans, huge trees with crowns -go away thither, apsarases; you have been recognized! Where your swings, golden and silvery, are, and where cymbals and lutes sound jointly -go away etc.
[…] 53 Cf. Joshi 1977, pp. 37-8, 48-9. In KGS 25.30, 35, describing the marriage ceremonies, the "maiden" (kanyā), is said to have sacrificed to Aryaman, the old Vedic god of marriage, and to the gandharva-pativedana-, "gandharva knowing/finding the husband"; as the offerings to the two deities are described in more or less identical words, and closely following each other, they appear to fill similar functions. The maiden's words, "May that divine gandharva release us (me) from here; not from that dwelling" (25.35: so 'smān devo gandharvaḥ preto muñcātu māmuṣya gr̥ hebhyaḥ; similarly of Aryaman, 25.30), obviously are a prayer for leaving the parents' house for that of the husband. (Aryaman pativedana-is addressed with similar words in AV 14.1.27-8.) Of interest are also some words uttered by the husband in connection with the first sexual intercourse of the newly-married couple (variants in KGS 29.1; Hiraṇyakeśi-Gṛhyasūtra 1.7.24.6): "The (magic charm of) concord that belongs to the cakravāka birds, that is brought out of the rivers, of which the divine Gandharva is possessed, thereby we are concordant" (transl. Oldenberg 1886, vol. 2 p. 198 [transliteration modernized]: cākravākaṃ saṃvananaṃ yan nadībhya udāhr̥ tam / yad yukto devagandharvas tena saṃvaninau svake). In all these texts, the gandharva is invoked to grant a (happy) marriage. Finally, the Āgniveśya Gṛhyasūtra, 2.5.61ff, includes a nuptial litany in which various Vedic deities and forces are identified with the gandharva (sing.) and his apsarases. 54 The gandharvas here are mentioned together with rákṣases (1-2) and piśācás (10); likewise in AV 11.9.16, 12.1.50. Kuiper (1996, p. 246 ) sees the "original" gandharva as a comparatively benign creature, and holds the lofty ṛgvedic gandharva (see below) to be an older conception than the menacing being seen in the AV; the possibility that this contrast has to do with different "genres" is considered, but rejected. Instead, Kuiper suggests "that a foreign influence has contributed to a 'demonization' of the Gandharvas, in that a group of non-Aryan demons was grafted upon the Gandharva", but concedes that "this cannot well be connected with the fact that Rig-vedic references to Indra's slaying the Gandharva […] are only found in the Kaṇva book VIII" (ibid.). Judging from the Iranian parallels (for which see below), this might actually be the oldest conception of the gandharva. Personally, I believe that the gandharva was an ambiguous creature -neither god nor asura -who could appear in a positive function (as an intermediary between heaven and earth), but also carried darker characteristics (lustfulness, jealousy; causing insanity or miscarriage). 55 Kuiper's statement (1996, p. 253), "It is obvious that this charm was pronounced three days after the wedding, when the Gandharva(s) had to be expelled," cannot be accepted in consideration of what we now know about the development of this practice (I refer to the study by Slaje). Perhaps Wijesekera (1994a, p. 190) is more correct in stating that "the reference is to the gandharva's power of 'possessing' human beings and causing madness rather than to their general connection with women".
Of the hither-dancing gandharva with plaited hair, husband of the apsarases, do I crush the testicles, I tear off (?) the penis … One like a dog, another like a monkey, a boy all hairy -having become pleasant to behold, the gandharva goes after women. Him do we make disappear from here with a potent spell. The apsarases are your wives; you gandharvas are their husbands. Hurry off, immortal ones; don't go after mortals! 56 The notion that the gandharvas are by nature ugly, but take on beautiful forms in order to seduce women, is very unusual; elsewhere, only their famed beauty is referred to. 57 The gandharvas and apsarases, dwelling in trees, are also implored not to harm a wedding procession, and specifically not the bride (AV 14.2.9). The lascivious nature of these beings is a cause of fear, but is also befitting deities of fertility and procreation. There is, thus, no actual contradiction when the Śāṅkhāyana Gṛhyasūtra 1.19.1-2, describing the sacrament for begetting offspring, implores Viśvāvasu to go away (with the verse RV 10.85.21, quoted above), only to invoke him again as the actual act is about to take place: "You are the mouth of the gandharva Viśvāvasu", says the husband as he touches the wife's private parts. While the gandharva's desire for the wife is feared, his powers are nonetheless invoked (as it seems) for successful procreation. 58 The lust for women is a common theme in Middle Vedic 56 nadīṃ yantv apsaráso 'pā́ṃ tārám avaśvasám/ gulgulū́ḥ pīĺā nalady àukṣágandhiḥ pramandanī/ tát páretāpsarasaḥ prátibuddhā abhūtana// yátrāśvatthā́ nyagródhā mahāvr̥ kṣā́ḥ śikhaṇḍínaḥ/ tát páretāpsarasaḥ prátibuddhā abhūtana// yátra vaḥ preṅkhā́ háritā árjunā utá yátrāghātā́ḥ karkaryàḥ saṃvádanti/ tát páretāpsarasaḥ prátibuddhā abhūtana// […} ānr̥tyataḥ śikhaṇḍíno gandharvásyāpsarāpatéḥ/ bhinádmi muṣkā́v ápi yāmi śépaḥ// […] śvéváikaḥ kapír iváikaḥ kumāráḥ sarvakeśakáḥ/ priyó dr̥ śá iva bhūtvā́ gandharváḥ sacate stríyas/ tám itó nāśayāmasi bráhmaṇā vīryā̀vatā //jāyā́ íd vo apsaráso gándharvāḥ pátayo yuyám/ ápa dhāvatāmartyā mártyān mā́ sacadhvam//. (4.37.2-5, 7, 11-12.) 57 For example, the gandharvas and apsarases are represented at the horse sacrifice by "beautiful" (śobhanāḥ) young boys and girls (ŚB 13.4.3.7-8); they are elsewhere associated with gandha-, "fragrance" (due to etymologizing), and rūpa-, "shape, beauty" (9.4.1.4; 10.5.2.25; cf. also Mbh 5.44.15), "whence if any one goes to his mate he cultivates sweet scent and a beautiful appearance" (9.4.1.4, tr. Eggeling). In the epics and in early Buddhist texts, the beauty of the gandharvas is a common motif; the handsomeness of the gandharva king Citraratha is proverbial (cf. Hopkins 1915, pp. 156-57). Their fondness of garlands, ornaments and fine clothes (ibid.) is notable, and may be traced to Vedic beliefs; cf. TĀ 3.10.3 (gandharvāpsarābhyaḥ sragalaṃkaraṇe) and perhaps already the celestial gandharva of RV 10.123.7, wielding bright weapons and clothing himself in a scented garment "to look like the sun" (dr̥ śé káṃ svàr ṇá). Apparently, the gandharvas and apsarases have elaborate hairdos, a topic which will be discussed later. Lastly, it may be mentioned that the gandharvas are quite often likened to the sun in splendor; besides RV 10.123.7, we have a "sun-hued" (sū́ryatvag) gandharva in AV 2.2.2; a gandharva named Sūryavarcas ("having the splendor of the sun") is mentioned in Baudhāyana Śrautasūtra 18.46 and TĀ 1.9.3, and in the Mbh (Hopkins,p. 153); the gandharvas Citraratha and Vasuruci share the patronymic Sauryavarcasa-in AV 8.10.27, and the wedding hymn of KGS has (v. 14) an apsaras Sūryavarcasinī, surely to be identified with Suriyavaccasā, daughter of the gandhabba Timbarū and lover of Pañcasikha, in the Dīgha Nikāya 2.263. When speaking of her five "gandharva husbands" (see below), Draupadī repeatedly uses the adjective sūryavarcasa-for them (Mbh 4.15.33;21.15), suggesting that this was a standing description of gandharvas. 58 If, as the commentarial literature claims, the gandha-smeared staff represents the gandharva even in texts where Viśvāvasu is not mentioned, then the procedure laid down in BGS 1.5.17ff is of great interest; after the husband and wife have spent three nights in chastity with the staff placed between them, the husband lifts it up with the words, "From nourishment, from the earth are you sprung, o Tree; grow with a hundred shoots! …" (ūrjaḥ pr̥ thivyā adhyutthito 'si vanaspate śatavalśo viroha). He then hands it over to the wife with the words, "I unite you with offspring, like the surā-drink with māsara-!" (prajayā tvā saṃsr̥ jāmi māsareṇa surām iva). The wife takes it, saying, "May I have offspring!" (prajāvatī bhūyāsam); then hands it back to the husband with words wishing him to obtain offspring and cattle (a wish repeated by the husband). It is obvious that the staff is invested with powers of fertility; it is, furthermore, to be made of udumbara-wood, which in Vedic ritual is frequently connected with fertility and abundance. Cf. also the commentaries on Āpastamba Gṛhyasūtra 8.9, where Haradatta states that the staff should be made from "a tree rich in sap" (kṣīrivṛkṣodbhavaḥ), while literature, where the gandharvas are given the epithet strīkāma-(to be discussed later on). In the law-texts of later times (the Dharma Sūtras and -Śāstras), the "gandharva marriage" appears in the canonical list of eight forms of marriage, as a union based on love or desire, disregarding societal conventions; it therefore belongs to the four "lower" forms of marriage, which are forbidden to brahmins. 59 AV 8.6 presents us with a much more harmful aspect of the gandharvas in relation to offspring. This is a spell against creatures causing miscarriage. In 8.6.19, at least some of them are identified as gandharvas: They who […] cause the [new-]born ones to die; who lie next to the birthgivers -may the Yellowish one (a talisman?) drive the women-enjoying 60 gandharvas away, like the wind a cloud. 61 The direct cause of the miscarriage is the cohabitation of the gandharvas with the pregnant woman; their lust for women may thus have sinister consequences. 62 While demons causing miscarriage are a common feature in Vedic as well as later Indian mythology, 63 the explicit reference to the gandharvas as strībhāga-, enjoying or partaking of women, definitely connects the passage under discussion to the general mythology around gandharvas, who are, in somewhat later Vedic texts, described as strīkāma-. The potentially harmful nature of their relationships with mortal women 64 is also made clear in these texts, where, as we shall see, the gandharvas often appear as possessing women. In this particular context, however, it can Sudarśana specifies that the wood should be from a banyan, an udumbara, an aśvattha or a plakṣa, since "these are the homes of the gandharvas and apsarases" (ete vai gandharvāpsarasāṃ gṛhāḥ; quoting TS 3.4.8.4). 59 Cf. e.g. Kane 1974, pp. 516-523. 60 Lit. "sharing in women". 61 yé amnó jatā́n māráyanti sū́tikā anuśérate/ strībhāgān piṅgó gandharvā́n vā́to abhrám ivājatu//. 62 This is not surprising, considering the often dangerous influence of similar semi-divine beings; cf. Winternitz 1895, pp. 154-55, who, commenting on Nejameṣa, notes "how closely connected the two ideas are of a deity dangerous to children, and a deity helpful in the procreation of children". 63 Note already the short hymn RV 10.162, imploring Agni the Slayer of Demons (rakṣohan-) to chase away an incubus hurting the embryo inside the womb; the sexual aspect is brought out clearly in v. 4: "Who separates your thighs, [who] lies between the married couple; who licks the inside of the womb -him do we make disappear from here." (yás ta ūrū́ viháraty antarā́ dámpatī śáye / yóniṃ yó antár āréḷhi tám itó nāśayāmasi //) In 5-6, the demon is said to lie down with the woman, having assumed the form of her brother, husband, or lover, or after having overwhelmed her with sleep. Cf. also ŚB 3.2.1.40 (tr. Eggeling): "And as to his saying, 'Brâhman,' uncertain, as it were, is his origin heretofore; for the Rakshas, they say, pursue women here on earth, and so the Rakshas implant their seed therein. But he, forsooth, is truly born, who is born of the Brahman (neut.), of the sacrifice …" (atha yad brāhmaṇa ityāha, anaddheva vā asyātā purā jānam bhavatīdaṃ hy āhū rakṣāṃsi yoṣitam anusacante tad uta rakṣāṃsy eva reta ādadhatīty athātrāddhā jāyate yo brahmaṇo yo yajñāj jāyate.) 64 Kuiper (1996, p. 243) rejects the notion that the gandharva's claims to the bride were originally thought of in negative terms. The bride, like the soma (see below), was under the guardianship of the gandharva while in a sort of "quarantine", required to divest her of potentially dangerous influences. Afterwards, the gandharva was simply asked to leave, without any threat or force being necessary. This is the picture gained from RV 10.85, where Viśvāvasu is said to have possessed the bride after Soma but before Agni, and is later implored to leave her to her husband and seek out another, unmarried girl. The later conception "may have had its origin in feelings of frustration (not to say Freudian castration) on the part of the husband" (ibid.) while his bride was in the care of someone else. Kuiper is probably right as far as the "quarantine" is concerned, though the three-day period is, as we have seen, a younger custom. I do not, however, think that the dismissal of the gandharva must "originally" have been so unproblematic as Kuiper claims; the gandharva's desire for women is already a prominent motif throughout the AV, the oldest parts of which cannot be much later than RV 10.85, and though the lasciviousness of this kind of being is only hinted at in the RV (e.g. 10.123.5, quoted below), this is so prominent a characteristic of the gandharvas from the AV onwards that it is hard to see how it could have been derived simply from a reinterpretation of the marriage custom (as Kuiper suggests), rather than vice versa. also be seen as a negative side of their power over procreation; 65 note that the gandharvas and apsarases, according to the PB passage quoted earlier, preside over a person's "offspring and childlessness".
Almost all the texts discussed so far belong to post-ṛgvedic times. An outline of the historical development of this mythological being will, of course, not be complete without taking into account the references in the earliest Veda, however sparse, scattered, and obscure these may be. It should be made clear already here that the conception of the gandharvas in the RV seems to differ in some respects from that presented in the AV and Middle Vedic texts. This is, no doubt, partly due to changes or developments in the later conception -this, I suspect, is particularly true as regards the gandharvas' role in rituals -but at the same time, the nature and style of the ṛgvedic hymns may account for some of the differences; and as the references to the gandharva in this Veda are sparse, too much should perhaps not be made of its silence concerning some typical traits of the gandharva (many of which do appear already in the AV).

gandharvá-in the Ṛgveda
We have neither the space, nor the need, to give an exhaustive survey of the earliest evidence, as this has already been studied in detail several times. 66 The most prominent traits may be summarized as follows: the gandharvá-(almost always sing.) of the Ṛgveda is a celestial creature, with a somewhat unclear relationship to the gods, often appearing in the company of his lovers, the apsarases; his abode is in the waters (ap-), which are often specified as the heavenly waters -there is thus no discrepancy between passages locating the gandharva in the heavens (e.g. 9.85.12 and 10.123.7: "the gandharva stood upright upon the firmament" 67 ) and those that simply speak of "the gandharva of/in the waters". Some later texts have the gandharvas and apsarases residing in trees (AV 4.37.4; 14.2.9; TS 3.4.8.4 68 ), like the yakṣas of post-Vedic religion, or (as for the apsarases) in rivers (AV 4.37.3; JB 1.42, 44 69 ); and this might reflect more popular beliefs. The antiquity of the "gandharva of/in the waters" (9.86.36; 10.10.4) is, however, clear from parallels in ancient Iranian texts; here we find (Yašt 5.38;15.28;19.41) the gandarəßa-as a monster inhabiting the celestial sea Vourukaša, where it is eventually slain by the hero Kərəsāspa. Traces of some similar myth have been seen in RV 8.1.11 and 8.77.5, where Indra is said to have defeated and "pierced" (abhí … atṛṇad) the gandharva under unclear circumstances. This is in striking contrast to the hymn 10.139, where the gandharva appears as Indra's helper, aiding him (according to Lüders' interpretation 70 ) in his search for the sun that was lost in the waters. The apparently demoniacal nature of the gandharva in book 8 being in contrast to his rather lofty appearance elsewhere in the RV, Iranian influence can perhaps not be entirely ruled out. 71 The Iranian evidence does, however, confirm that there was originally only one gandharva; the plural of the word being, with a single exception, 72 found only in the younger parts of the RV, and only twice. 73 Though it is of slight relevance to the present study, it may be noted that the word gandharva--and thus, it seems, the being designated by it -is usually considered a substrate word in  This assumption would be supported by the fact that the word is not found in any of the "family books" of the RV (with the single exception of 3.38.6), thus pointing, as Hillebrandt noted, 75 to a late acceptance of this kind of being into the brahmanical religion. While Indian tradition, from the earliest times, has derived the word from gandha-"fragrance", 76 and the beings are said to subsist on the mere fragrance of plants and herbs, 77 this has commonly been rejected as a pseudo-etymology; indeed, "fragrance" does not occupy a prominent place in the lore around the gandharvas, and the notion of their feeding on fragrance (as well as the part played by scents in their ritual, as seen above) is perhaps more likely to have been derived from the "etymology". 78 Of doubtful derivation is also the word apsaras, denoting the female partners of the gandharvas. The meaning "moving (sṛ-) in the waters (ap-)", 79 though in keeping with their nature as nymphs dwelling in rivers and other bodies of water, is now usually rejected in favor of the reading a-psaras, "shameless", 80 apparently referring to their promiscuous character. 70 Lüders 1959, pp. 537-41. 71 Oberlies (ibid., p. 100) notes in this regard that the eighth book of the RV, in which both passages are found, has been considered as having a more westerly origin than other books; cf. also Hillebrandt 1999, p. 252. Significantly, RV's book 8 was composed by the Kāṇvas and Āṅgirasas, who are also (as is now generally recognized; cf. Parpola 2015, pp. 131-33) the poets behind the Atharvaveda, where the gandharva's demoniac nature reminds one of RV 8.1.11 and 8.77.5. A reminiscence of a connection with the atharvavedic seers may be seen in the BĀU 3.3.1 and 3.7.1, where two gandharvas who have possessed women identify themselves as "Sudhanvan Āṅgirasa" and "Kabandha Ātharvaṇa", respectively. 72 RV 3.38.6c-d, in which the poet addresses some unnamed gods with the words, "Having come here with my mind, I saw even the wind-haired Gandharvas under your commandment" (transl. Brereton and Jamison; ápaśyam átra mánasā jaganvā́n vraté gandharvā́m ̐ ápi vāyúkeśān). 73 In 9.113. 3 and 10.136.6. Cf. Hillebrandt 1999, pp. 252-53. 74 Thus Kuiper (1996, pp. 225-26): "Since an interchange v/b is excluded in words of IE origin, the different names [gandharva-and gaṇdərəßa-] point to a foreign (that is, non-Indo-European) origin … The interchange of a phoneme /ṛ/ with ar in Indo-Aryan cannot represent an IE ablaut, nor is /ṛ/ likely to have occurred in a non-Indo-European language. The name that was adopted into Proto-Indo-Iranian may have been *G(h)andh(a)rba-/*G(h)andh(a)rwa-or, if Skt dh is due to popular etymology, *gand(a)rb/wa-." Witzel (2003, pp. 39, 55) considers the word as part of a Central Asian linguistic substrate in Indo-Iranian, comparing it with other terms having the "suffix" *-arwa-, "which is seen only in religious terms"; cf. Vedic atharvan-, Śarva-. Cf. also Witzel 2004, esp. p. 605 n. 41, pp. 615, 620ff. 75 Hillebrandt 1999 Probably already in RV 10.123.7, where the heavenly gandharva clothes himself in a scented (surabhi-) garment. 77 Thus AV 8.10.27, 12.1.23; cf. ŚB 9.4.1.4, JUB 3.25.4. The Buddhist Saṃyutta Nikāya, 3.250, has the gandhabbas inhabiting "fragrant" roots, leaves, and other parts of trees. 78 Cf. however Mayrhofer, s.v., who considers the possibility of a derivation from *gandhas-/*gandhar-. 79 This is the etymology given in the Nirukta (5.13); the same understanding of the word seems to underlie AV 2.2.3c-d, where the apsarases are described as moving to and fro in the ocean (samudrá āsāṃ sádanaṃ ma āhur yátaḥ sadyá ā́ ca párā ca yánti). 80 Cf. e.g. Mayrhofer, s.v. (who also considers the meaning "shapeless"). Whether they were "originally" the spouses of the gandharvas (as they are already in RV), or have been paired with them later due to their similar characteristics, is a question of no greater bearing to this study; while the late passage 10.10.4 (cf. 10.11.2) speaks of "the gandharva in the waters and the water-maiden" (gandharvó apsv ápyā ca yóṣā) without explicitly calling the "water-maiden" an apsaras (she is referred to as gandharvī-in 10.11.2), it seems precocious to infer that the gandharvī-was the "original" spouse of the single gandharva. 81 In my view, the characteristics shared by gandharvas and apsarases already in the RV -among which are, as we shall see, beauty, eroticism, dwelling in water -are too prominent to permit the conclusion that these groups of beings are of separate origins. Kuiper 82 saw the "gandharva of the waters and the water-maiden" of 10.10.4 as the original conception of the single gandharva, living with his wife in the primordial waters, and believed that this conception survived in the legend (to be discussed later) told in JB 1.125-27, where we meet with a gandharva and his spouse -what kind of being she is we are not told -in their home floating on the waters. While not contesting that there was originally only one gandharva, I am not fully convinced by these isolated passages -one occurring in the very latest stratum of the RV, the other in one of the younger Brāhmaṇas. In the AV and some later texts, we find a single gandharva in a polygamous or even promiscuous relationship with several apsarases; 83 in RV 10.123.5, quoted below, we do indeed find the heavenly gandharva accompanied by a single apsaras, but he is here described as her "lover" (jārá-), perhaps suggesting something like the promiscuous relationships of these beings as known from later texts. This is perhaps the only ṛgvedic passage where the erotic associations of the gandharva, well known from the AV onwards, appear clearly. Otherwise, there are notable differences in the conceptions of the lofty, celestial gandharva of the RV and of the semi-demoniac being of later texts. Its incubus nature is implied only in the late wedding hymn 10.85, which is also one of the few ṛgvedic references to its potentially harmful or menacing nature. Conceivably, the two references in book 8 to a gandharva-demon defeated by Indra may point to a conception of this being more in keeping with that found in the AV and later texts. The tendency of the ṛgvedic singers to elevate deities and spirits by associating them with celestial gods like Soma or the Sun, might be responsible for the very different conception of the gandharva which appears in several hymns. Passages like 3.38.6 (the only mention of the gandharvas in one of the family books), where we encounter the "wind-haired gandharvas" under the command of some unnamed gods, and 10.136.6, describing the journey of a flying shaman along "the course of the apsarases and gandharvas and the wild beasts" (apsarásāṃ gandharvā́ṇām mr̥ gā́ṇāṃ cáraṇe), seem to imply a parallel, perhaps more popular, conception of these beings as terrestrial or atmospheric spirits. However, the ṛgvedic references to the gandharva(s) are scanty and often cryptic, and probably do not allow us to paint a complete picture of the earliest conception of these beings.
A few passages have been interpreted as implying a connection between the gandharva and the womb, or birth. These induced Pischel to consider the gandharva as identical with the fetus, and have more recently been treated by Haas (2004) in comparison with the Buddhist material. 84 Unfortunately, most of the relevant verses appear in (intentionally) obscure hymns of riddle-like poetry, and their exact import is far from clear. It should also be noted that words like garbhá-("womb" or "fetus") and yóni-("womb") occur frequently in this kind of hymns, usually in a mystical sense as referring to the divine "births" of Agni, Soma, or the sun.
RV 10.123.5 refers to the gandharva and apsaras in a kind of love-play: The apsaras, the young woman, smiling towards her lover, bears him in the highest heaven. Moving inside the dear one's womb, being dear, this Vena sat down on the golden wing. 85 The "lover" of the apsaras is the gandharva, being mentioned in the immediately preceding 10.123.4d ("the gandharva knew the immortal names" 86 ) and again in v. 7 ("The gandharva stood upright upon the firmament, wielding his bright weapons, turned hitherwards" 87 ). The notion that he moves inside her womb may refer either to sexual activity or to pregnancy; the latter interpretation seems to be supported by the word bibharti "bears"; the verbal root bharbeing often used to denote pregnancy or birth-giving. Possibly, the reference to the gandharva moving in his lover's womb may have an intentional double meaning, in which case the verse would give expression to a kind of mysticism with the gandharva being both the lover and the child of the apsaras. In any case, it does refer to the lustful nature of the gandharva(s), which elsewhere seems implied only in 10.85. Another difficult passage is RV 9.83.4. The immediately preceding half-verse (9.83.3c-d) runs as follows: "The magicians have measured it out through magic; the fathers, watching over mankind, have placed the germ (gárbha-)." 88 Then follows verse 4: "The gandharva guards its place here; the awesome one protects the generations of gods. With his noose the 84 This latter publication consists of a translation and study of all the RV passages mentioning the gandharva(s), in the light of Sāyaṇa's medieval commentary; this choice of method is of course open to criticism, and the author devotes some space to defend it. Hardly any attention is given to AV or to Middle or Late Vedic texts. The ultimate purpose of the work is to demonstrate that the gandharva of the RV is an atmospheric "Zwischenzustand"-creature, mediating between heaven and earth, and thus an immediate precursor of the much later concept of the gandhabba, as (in the interpretation embraced by the author) a deceased spirit waiting to be reborn. Haas further attempts to read seeds of the later transmigration doctrine into the passages concerned. While any study of the elusive ṛgvedic gandharva should be welcome, there are obvious problems with the author's choice of method (cf. the wholly negative review by Jamison [2008]), in adducing late Buddhist texts and medieval exegesis to illuminate the RV passages, while passing over the later Vedic literature. The mediating character of the gandharva is, as we shall see, probable enough, but the passages studied by Haas are often enigmatic enough for an interpretation placing the gandharva in any kind of "middle" position to be, more or less forcefully, applied. (Cf. Jamison,p. 395: "[Haas'] other means of handling the problem is to see all sorts of "in-between" positions and states in the Rigvedic Gandharva passages, and then to argue that any kind of "zwischen" is equivalent to the Zwischenzustand between death and rebirth. Thus, sunrise, the production of words from thoughts, the freeing of the Vala cows, the Gandharva's role as intermediate bridegroom in the wedding hymn, and name-giving can all be used as evidence for the Rigvedic origins of the later, specifically Buddhist, function of the gandharva.") 85 apsarā́ jārám upasiṣmiyāṇā́ yóṣā bibharti paramé vyòman/ cárat priyásya yóniṣu priyáḥ sán sīdat pakṣé hiraṇyáye sá venáḥ//. 86 vidád gandharvó amŕ̥ tāni nā́ma. 87 ūrdhvó gandharvó ádhi nā́ke asthāt pratyáṅ citrā́ bíbhrad asyā́yudhāni. 88 māyāvíno mamire asya māyáyā nr̥ cákṣasaḥ pitáro gárbham ā́ dadhuḥ//. noose-lord seizes the foe; the supreme well-doers have consumed the honey." 89 The hymn is, like all others in RV:s book 9, dedicated to Soma, and the "honey" is, as usual, the somadrink, apparently consumed by the pious dead in heaven. As Kuiper and Oberlies have shown, and as we will discuss later on, the ṛgvedic gandharva is the guardian of the soma in heaven; and it is obviously in this function he appears in this verse. But the preceding, somewhat less clear, verse may indicate yet another function; here, the soma is, as often, likened to a germ or fetus (gárbha-). May the immediately following mention of the gandharva indicate a role as protector of the fetus or embryo in the womb? The mention of the gandharva's protecting the generations or births (janiman-) of the gods would perhaps support such an interpretation, though its meaning is not very clear. If the proposed interpretation is correct, however, the gandharva's association with the fetus, as well as its guardianship of the soma, may both be expressed in these verses through the poetical designation of soma as a "fetus".
A connection with the womb is again met with in RV 10.177.2. I quote here the first two verses of this short (three verses) but obscure poem: The bird, anointed with the magic of the asura, the discerning ones behold with their heart, with their mind. Inside the ocean the seers discern [it]; the masters long for the place of the sun-rays. The bird carries Speech in its mind; the gandharva proclaimed it inside the womb. That flashing, sun-like inspiration the seers guard in the place of Truth. 90 "The bird" is presumably the sun (thus Sāyaṇa) in the ocean of heaven; here appearing in a mystical sense as connected with esoteric insight (Geldner: "das innere Licht der seherischen Erkenntnis und Erleuchtung im Herzen"). The gandharva is mentioned as having proclaimed (sacred) Speech to it, "inside the womb". Whatever the symbolic import of all this, the gandharva here, again, appears to be connected with the womb.
Some similar conception may be the basis of a passage in the oft-discussed dialogue-hymn 10.10. In an attempt to make her twin-brother, Yama, agree to an incestuous relationship, Yamī appeals to divine will, stating that the gods have preordained this incestuous union. In v. 4, Yama counters her arguments; the second half-verse runs thusly: "The gandharva in the waters and the water-maiden -that's our origin [lit. 'navel'], that's our ultimate kinshipbond." 91 The meaning of these words has been the subject of much discussion. The mention of the "kinship-bond" (jāmi-) and the nābhi-, meaning "navel" but often used in the sense of origin or connection through birth (Jamison and Brereton: "umbilical tie"), has led a number of scholars to the conclusion that the gandharva and the "water-maiden" -clearly an apsaras 92 -are the parents of Yama and Yamī. Such a conclusion, however, has some difficult implications, as Yama's (and therefore, presumably, Yamī's) parents are elsewhere (RV 10.17.1ff,etc.) said to be the semi-divine Vivasvant and Saraṇyū, daughter of the god Tvaṣṭṛ; the patronymic Vaivasvata is used of Yama even in post-Vedic tradition. The 89 gandharvá itthā́ padám asya rakṣati pā́ti devā́nāṃ jánimāny ádbhutaḥ/ gr̥ bhṇā́ti ripúṃ nidháyā nidhā́patiḥ sukŕ̥ ttamā mádhuno bhakṣám āśata//. 90 pataṃgám aktám ásurasya māyáyā hr̥ dā́ paśyanti mánasā vipaścítaḥ/ samudré antáḥ kaváyo ví cakṣate márīcīnām padám ichanti vedhásaḥ// pataṃgó vā́cam mánasā bibharti tā́ṃ gandharvò 'vadad gárbhe antáḥ/ tā́ṃ dyótamānāṃ svaryàm manīṣā́m r̥ tásya padé kaváyo ní pānti//. (10.177.1-2.) 91 gandharvó apsv ápyā ca yóṣā sā́ no nā́bhiḥ paramáṃ jāmí tán nau. 92 Cf. the ápyā … yóṣaṇā of the following hymn (10.11.2), who is called a gandharvī-. evidence being unanimous on this point, the question is what to do with the gandharva and the water-maiden of RV 10.10.4. Some scholars have tried to solve it all by simply identifying the gandharva as Vivasvant. 93 However, Vivasvant is nowhere else referred to as a gandharva, or even put in connection with these beings; nor is Saraṇyū ever called an apsaras, or a "water-maiden". Unless we assume an alternative, otherwise unattested, tradition, according to which Yama and his sister were begotten by a gandharva and an apsaras, we must therefore conclude that the verse in question does not refer to their parentage.
Of what kind, then, is their relationship to these beings? Schneider, after refuting the parenthood-theory, proposes, with some caution, "bei Gandharva und die Wasserfrau an eine weiter zurückliegende Abkunft zu denken, etwa an das Urpaar eines clans der Amṛtas [immortals; gods], für den das Gebot der Exogamie … gilt, das ja, religionssoziologisch betrachtet, nicht vom Verbot des Inzests zu trennen ist" 94 . Yama's argument against the proposals of incest would, then, refer back to the "immortals", mentioned by Yamī in the preceding verse (v. 3) as wishing for the intercourse to take place. According to Schneider's interpretation, Yama would thus counter his sister's argument by referring to the will of even higher immortals. This explanation of the passage differs from some others in actually considering what relevance Yama's invoking of the (will of the) gandharva and the maiden may have for his argumentation: these beings are, clearly, of another kind than the "immortals" invoked by Yamī, and constitute the "ultimate" kin of the twins. The same kind of interpretation has been proposed by Kuiper,95 who takes paramá-to mean that the gandharva and his consort are the primordial ancestors of Yama and Yamī, as well as the gods.
This is certainly possible. But the idea of a gandharva and an apsaras constituting some sort of "Urpaar" or being older than the gods does not, to my knowledge, find any support either in the RV or in younger texts. Rather, the true nature of the gandharva and the water-maiden may be explained in the light of the first half of the next verse (v. 5 a-b), in which Yamī counters her brother's argument: "In the womb, the Begetter made us husband and wife -god Tvaṣṭṛ, the Impeller, possessor of all forms." 96 Tvaṣṭṛ is, as is well known, the artisan of the gods, who among other things is credited with shaping the embryo in the womb (cf., for 93 E.g. Barnett 1928;Wijesekera 1994a, p. 187;Haas 2004, p. 137 (following Sāyaṇa who, indeed, makes this identification). 94 Schneider 1967, p. 15. 95 Kuiper 1996. He rejects the possibility that the conception of gandharvas as deities presiding over offspring might explain the passage under discussion, claiming that, "Yama's appeal to the Gandharva as his ultimate origin, while declining Yamī's proposal, is even a strong argument against the Gandharva's being a genius of procreation"; the latter notion is, according to Kuiper, not to be found in the oldest Vedic texts. He does not, however, discuss those RV passages which, however obscure, connect the gandharva with the womb or the embryo. In support of his belief that the gandharva and his consort constitute a primordial couple, more ancient than the gods, Kuiper depends exclusively on the much later JB 1.125-27 and 1.154-55 (to be discussed later); yet even these passages do not state that these beings are older than the gods, but only that they did not take part in the war between gods and asuras. -Oberlies (2009;2012, p. 144) sees the gandharva and the apsaras of RV 10.10.4 as presiding over procreation, but also seems to connect this function with their being primordial begetters: "the gandharva and the apsaras who dwell in the heavenly waters and near the sun beget the first human beings, who are believed to descend from the sun" (2009). Assuming that the sun here, as the begetter of mankind, is Vivasvant (cf. Oberlies 2012, pp. 218-19, 321), i.e., Yama's and Yamī's father, I am not sure who begets whom according to this reconstruction. 96 gárbhe nú nau janitā́ dámpatī kar devás tváṣṭā savitā́ viśvárūpaḥ/.
instance, RV 10.184.1); this activity of his usually being referred to as the shaping of "forms" (rūpa-) -thus his epithet Viśvarūpa-, "(possessor of) all forms". 97 The argument, which is as clear as it gets, is that the shaper of embryos intended the incest of the twins to take place already when shaping them inside the womb. It seems likely that the gandharva and the watermaiden stand for some similar, but not identical, activity; for instance, the placing of the embryo inside the womb. Yama's argument against incestuous intercourse would, then, be something like this: while Yamī invokes the will of the immortals -among which the twins, as Schneider points out, belong -her brother counters this argument by pointing beyond their own generation, or that of their parents; declaring the gandharva and the water-maiden to be their ultimate (paramá-) kin and origin, having brought them as embryos to the womb, perhaps from some supra-terrestrial sphere like the heavenly waters. This may not necessarily mean that these two beings are opposed to incest, but rather nullifies the authority -through kinship -of the immortals. (Possibly, however, there is an implicit reference to the sexual habits of the gandharvas and apsarases, which are clearly not incestuous but rather, as we shall see, promiscuous.) Yamī, in her turn, counters this argument by invoking the will of the god Tvaṣṭṛ, whose part in their creation was similar to that of the gandharva and the watermaiden -he shaped them as embryos in the womb and, presumably, decided their destinies. This interpretation is, however, by no means certain. Wholly implausible is, in any case, Pischel's attempt to identify the gandharva here with the fetus. 98 Gandharvas, soma, women, and the notion of exchange between the worlds One important trait of the ṛgvedic gandharva is its connection with sóma-. 99 As Lüders 100 demonstrated, this sacred herb was thought to have its origin in the waters of heaven, i.e., in the abode of the gandharva; and indeed, we find the gandharva as guardian of the soma in the heavenly waters: here is the "fixed place of the gandharva" (gandharvásya dhruvé padé, 1.22.14), dripping with ghee and milk (i.e., probably soma); "here the gandharva guards his place" (gandharvá itthā́ padám asya rakṣati, 9.83.4), where the well-doers drink the "honey" (mádhu-; soma); 101 here is the "flood" (síndhu-) in the "highest heaven" (paramé vyòman), where "they lick the streams of ambrosial honey" (rihánti mádhvo amŕ̥ tasya vā́ṇīḥ) and the gandharva stands "upon the firmament" (ádhi nā́ke) (10.123.3-5, 7). In 9.113.3, the gandharvas (pl.) are mentioned as placing the juice (rása-) in the soma (plants), 102 apparently having brought it from the celestial waters. Oberlies has postulated an Indo-Iranian protomyth in which the gandharva, dwelling in the waters of heaven, was defeated by a god or hero 97 See in general Macdonell 1897, pp. 116-18. 98 Pischel & Geldner, pp. 78-9. 99 See especially Kuiper 1996;Oberlies 2005;Hillebrandt 1999, pp. 248ff; cf. also Barnett 1928. 100 Lüders 1951 For the gandharva as guardian of soma, see further Oberlies 2005. I do not think, however -like Oberlies and others before him -that Soma is "identified" with the gandharva in some passages. As Hillebrandt pointed out (1999, p. 249), gandharva-is several times used in the RV as an appellative for various other entities. When Soma in 9.78.3 is described as surrounded by the "apsarases of the ocean" (samudríyā apsaráso), it is a matter of mere comparison: the soma-stalk soaked in water is (by implication) likened to the gandharva amidst the apsarases in the heavenly waters (cf. AV 2.2). If 9.85.12 and 86.36 actually do identify Soma and the "heavenly gandharva", then these are isolated instances, and the identification an expression of typical priestly speculation. 102 parjányavr̥ ddham mahiṣáṃ táṃ sū́ryasya duhitā́bharat/ táṃ gandharvā́ḥ práty agr̥ bhṇan táṃ sóme rásam ā́dadhur índrāyendo pári srava //. in a struggle for the divine beverage (i.e., *sáuma-) guarded by it. 103 This seems likely enough, but it should be noted that, while the haoma-is indeed situated in the lake Vourukaša, it has no part in the gandarəßa-myth; nor does soma, in those passages which mention the gandharva's defeat at the hands of Indra. Moreover, the gandharva is, as Oberlies himself notes and as Kuiper was the first to emphasize, elsewhere in the RV clearly guarding the soma for the benefit of the gods. In fact, as late as in the Kauṣītaki Brāhmaṇa (12.4.2) we find the notion that "the Gandharvas as commissioners in the waters guard the Soma of Indra" 104 (transl. Keith). 105 This raises some questions as to the exact relationship between the gandharva(s) and the gods. In Middle Vedic texts we do find a myth -the most common one which involves the gandharvas -in which the soma is wrested from the gandharvas by the gods by means of trickery. This myth appears in several texts, but its plot is basically the same, except for a few variants: 106 the gandharvas barter the soma in exchange for the goddess Speech (Vāc), the personification of Vedic sacred utterances. It usually begins with the gods obtaining the soma from heaven, only to have it stolen from them by the gandharva Viśvāvasu. 107 This might seem strange if the view is accepted that the gandharvas originally guarded the soma for the gods; conceivably, the nature of these beings as intermediaries (for which see below), holding the soma between its descent from the highest heaven and before its reaching the gods or humans, has played a part in the formation of the younger myth. Possibly the mention of the gandharvas as "seizing" (gṛbh-) the heaven-born soma in RV 9.113.3 may have contributed too.
I quote here the version of the myth found in the Śatapatha Brāhmaṇa. After the theft of the soma, the gods ponder on how to retrieve it: They said, 'The Gandharvas are fond of women: let us send Vāc (speech) to them, and she will return to us together with Soma.' They sent Vāc to them, and she returned to them together with Soma. The Gandharvas came after her and said, 'Soma (shall be) yours, and Vāc ours!' 'So be it!' said the gods; 'but if she would rather come hither, do not ye carry her off by force: let us woo her!' They accordingly wooed her. The Gandharvas recited the Vedas to her, saying, 'See how we know it, see how we know it!' The gods then created the lute and sat playing and singing, saying, 'Thus we will sing to thee, thus we will amuse thee!' She turned to the gods; but, in truth, she turned to them vainly, since she turned away from those, engaged in praising and praying, to dance and song. Wherefore even to this day women are given to vain things: for it was on this wise that Vāc turned thereto, and other women do as she did. And hence it is to him who dances and sings that they most readily take a fancy. 108 (Transl. Eggeling; transliteration modernized.) There is an intended irony in this story which seems to have passed unnoticed in most discussions: 109 the usual roles of the gandharvas and the gods, respectively, are temporarily switched. It is, usually, the gandharvas who are occupied with dancing, singing and music; in post-Vedic mythology (Buddhist; epics, etc.) we find them typically as musicians of heaven (cf. gāndharvaveda-, the art of music), and their wives, the apsarases, as celestial dancers. 110 But already in the Atharvaveda (4.37.7) there appears a menacing, "hither-dancing, plaithaired" (ānr̥tyataḥ … śikhaṇḍíno) gandharva, and the gandharvas and apsarases in general are said to "revel in feasting" (sadhamā́daṃ mad-; 7.109. 3,5; 14.2.34; cf. 4.34.3; 4.38.3 [dancing apsarases]). This is a constant feature in later literature, which associates them with frolicking, games, and coquetry; the "swings" referred to in AV 4.37 recur in PB 12.11.10, where the gandharva Ūrṇāyu is introduced while "swinging amidst the apsarases" (apsarasāṃ madhye preṅkhayamāṇam), and in JUB 3.25 the gandharvas are connected with the attributes fragrance (gandha-), joy (moda-), and pleasure (pramoda-), and the apsarases with laughter (hasa-), play (krīḍā), and sexual intercourse (mithuna-). This corresponds well to the nature of these beings as depicted in the epics and early Buddhist texts. 111 Believing, thus, that the goddess of sacred speech will not be enticed by such fleeting pleasures as dance and song, the gandharvas in the myth revert to reciting the Veda, laying off their old style; but the gods, understanding the mind of women, start imitating the gandharvas' old habits, and win over the goddess.
The ritualistic import of parts of the myth has long been recognized, and is indeed made explicit in the sources themselves. The cow used for (ritualistically) buying the soma-stalks is thus declared to be Vāc; and, before its preparation, the soma is left exposed for three days, etarhi moghasaṃhitā eva yoṣā evaṃ hi vāg upāvartata tām u hy anyā anu yoṣās tasmād ya eva nṛtyati yo gāyati tasminn evaitā nimiślatamā iva. 109 Ludvik (1998, pp. 348-49 n. 10), in discussing this myth, brings up the fact that "song is the domain of the gandharvas", but then concludes, "By the time of the Mahābhārata, the gandharvas are indeed musicians (see Hopkins 1915: 154), but this does not seem to have been the case in Vedic texts. At most, the AV (4.37.7ab) mentions the 'dancing' gandharva: ānr̥tyataḥ… gandharvásya…" While this may be true, gandharvas do appear as celestial singers fairly early; the gandharva Pañcasikha of the Buddhist Nikāyas is both a singer and a player of the vīṇā or lute (like the epic Nārada), and gandharvas carrying this instrument appear on the reliefs of Bharhut, Sikri, and Loriyan (3 rd to 1 st cent. BC.; cf. Hillebrandt 1987, p.184). This widespread conception, found in the brahmanical as well as Buddhist traditions, must be more ancient than the early Buddhist texts. The Vedic notion of the gandharvas and apsarases as constantly frolicking and feasting, furthermore, should probably not be separated from the conception of them as singers and dancers; note also that the apsarases are depicted in JB as indulging in "dance and song and the sound of the vīṇā" (nṛttagītaṃ vīṇāghoṣo, JB 1.42, 44), and their abode is in AV 4.37.5 said to be resounding with cymbals and lutes. On the other hand, I am not sure if the vīṇā in the myth is to be connected with that belonging to the goddess Sarasvatī (often identified with Vāc) in post-Vedic mythology (Ludvik,p. 357). 110 Hillebrandt 1987110 Hillebrandt (1906Wijesekera 1994a, pp. 192-93;Hopkins 1915, pp. 154ff. Although the singing and vīṇā-playing gandharva Pañcasikha of early Buddhist tradition (Dīgha Nikāya, etc.) is not (to my knowledge) found in Brahmanical texts, it is certainly interesting that an apsaras named Pañcacūḍā, "Five-Plaits" (the same meaning as pañca-śikha-) appears in the Mbh (Poona edition: 12.319.18;13.3.11,38.2ff,151.10), and apparently already in TS 5.3.7.2, which prescribes the laying down (on the fire altar) of a pañcacoḍā brick for winning the company of apsarases in the next world. Cf. also ŚB 8.6.1.11ff (the pañcacūḍā bricks and apsarases). The epithet śikhaṇḍin-of the gandharva in AV 4.37.7 indicates a similar hairstyle. The Sikhaṇḍin of Dīgha Nikāya 2.263, son of Sakka's charioteer Mātali, is apparently a gandhabba; he married Suriyavaccasā, daughter of the gandhabba Timbarū (Tumburu of Sanskrit texts), but later lost her to Pañcasikha. Elaborate hairdos may reflect the usual concern of these beings with beauty and sensuality; cf. also Vasilkov 1990, pp. 394-95, for some different suggestions. 111 Cf. Hopkins 1915, pp. 153ff;Bhattacharyya 2000, pp. 57-8, 123ff. during which it is said to be in the ward of the gandharvas. 112 Kuiper pointed out the parallel between the latter practice and the three nights during which the bride was (apparently) in the care of Viśvāvasu, and provided the following interpretation: "The two parallel cases confirm the conclusion that the Vedic 'Tobiasnächte' involved a kind of quarantine, which was required in order to divest the bride and Soma of their inauspicious nature." 113 This theory of a "quarantine" has been expanded on in a paper by Oberlies,114 who explains the main function of the Vedic gandharva as connected with transfer and mediation: besides guarding the soma in the heavenly waters, the gandharva is also responsible for transferring it to earth, as reflected in the ritual but also hinted at already in RV 9.113.3, where the gandharvas are said to have put the soma-juice (rasa-) in the plants (on earth): The buffalo, caused to grow by Parjanya, that the daughter of the sun has born -that one the gandharvas received and placed as sap in the soma. Flow, O Soma, for Indra! 115 The notion that the daughter of the sun "bore" (bhar-) the heavenly soma is interesting; this word often has the meaning "to be pregnant". There may, thus, be an intended analogy to the gandharvas' connection with embryos -the gandharvas place the sap in the soma plant, in the same way as the embryo in the womb.
The gandharvas in this passage appear in a sort of intermediary position; they possess the soma after the "daughter of the sun" (in heaven), but before its descent to earth. The case is similar with the bride: as for 10.85.40-41, where the bride is said to have belonged first to Soma, then to the gandharva, to the (nuptial) fire, and finally to the bridegroom. Oberlies points out that the gandharva here appears in a mediating position, possessing the bride before the marriage ceremonies, but after Soma (whatever his role here might be). He suggests that this is a recurring trait of the Vedic gandharva; the later notion of the bride being under the care of Viśvāvasu for three nights is thus explained with reference to the same idea: the bride, as a stranger coming from a different clan, must initially become divested of her "inauspicious nature" (Kuiper); the Gṛhyasūtras also prescribe rituals for neutralizing those of her "aspects" or "forms" (tanu-) which are harmful to the husband, as well as her evil eye (ghora-cakṣus-). The soma, believed to originate in the other world, must similarly be put to three days' "quarantine" under the care of the gandharvas. To this are also compared the three days between initiation and the beginning of study required for the Veda-student, as well as the three days of inactivity following a funeral; the newly initiated student still being a stranger in his teacher's house, like the bride in the husband's, and the family of a deceased person having to go through a kind of "quarantine" to get rid of their impurity. 116 This is an ingenious theory which would account for a great deal of the beliefs and practices surrounding the gandharvas. It has, however, aspects which are not unproblematic. The gandharva is nowhere in the source-texts mentioned in connection with the three days following the student's initiation, or the three days after a funeral. The three nights of chastity 112 Cf. Kuiper 1996, pp. 234ff. 113 Kuiper 1996. 114 Oberlies 2005. More generally in Oberlies 2009;2012, pp. 142-45. 115 parjányavr̥ ddham mahiṣáṃ táṃ sū́ryasya duhitā́bharat/ táṃ gandharvā́ḥ práty agr̥ bhṇan táṃ sóme rásam ā́dadhur índrāyendo pári srava//. 116 Oberlies op.cit.,p. 98. are, as Slaje has shown and as has been discussed above, a practice belonging to the latest strata of Vedic literature; and when it first appears, it is only in some texts put in connection with the gandharva. The staff in the bed is likewise a late element added to the practice; it is probable, if not certain, that the gandha-smeared cloth wrapped around it is an allusion to the gandharva. 117 Oberlies accepts Slaje's outline of the development of the "Tobias-nights", adding that, though these may be a late feature, their precursor -the three nights following the end of the wife's period, after which cohabitation may take place -must "zu den dreitägigen 'Quarantänen' gehören". 118 Even so, the precursor is not, either, associated with gandharvas. The only possible way to account for the three nights of chastity by means of Oberlies' theory would, then, be to explain them as a late development based on an ancient belief. This is probable enough; but it might also be that these nights are directly inspired by the "quarantine" of the soma. If that is the case, then there remains only one example of the three days' quarantine that can be directly connected with the gandharvas: that period during which the soma was believed to be under the care of these beings. This, in turn, could well be accounted for with reference to the gandharva's mythical function as guardian of the soma in heaven.
While the theory of the gandharva's intimate connection with the "quarantine"-periodhowever attractive -can be brought into question on some points, I do find convincing the characterization of gandharvas as beings responsible for various kinds of transfer. Similar suggestions have indeed been put forward, in less elaborated forms, by previous scholars. Barnett, for instance, suggested a parallel between the ṛgvedic gandharvas' placing the sap in the soma plants, and the later view making them responsible for successful procreation: "Both the waters and the Sōma are in the highest heaven […] thence the waters, divine life-saps, are brought to earth by Gandharvas and Apsarases, who therewith impregnate men, animals, and vegetation. The Gandharva was thus constantly travelling from heaven to earth (rájasō vimā́naḥ, RV. X, CXXXiX. 5) for the benefit of the world …" 119 Gonda, similarly, pointed to the gandharvas' function as mediators of sacred knowledge in the RV: There can be hardly any doubt that "the gandharva" is represented as a mediator between the divine secrets and the minds of men: revealing speech and stimulating dhīḥ [inspired thoughts] he disclosed to them what they did not know previously … The gist of the various statements of the character and activities of these deities is, in my opinion, that they are genii of conception and procreation, who keeping watch over the place of conception come as "lords of being" (AV. 2, 2, 1), dwelling in heaven, as "Wesenskeime" or "Seelenwesen", into touch with various divine beings and phenomena of a similar nature. As such they know and reveal the immortal (AV. 2, 1, 2) and the divine secret, found the immortal names (ṚV. 10, 123, 4, cf. 10, 139, 6; 177, 2), proclaim that highest abode that is in secret (AV. 2, 1, 2) and Vāc (Speech) in the womb. 120 This association with hidden knowledge from the other world has also been pointed out by Kuiper,121 and, following him, by Oberlies; 122 it is found in the RV and AV as well as in 117 May this have been inspired by the gandharva's "clothing himself in a scented garment" in RV 10.123.7? But as mentioned in a previous note, the connection with both perfume and garments is common in Vedic and later literature. 118 Ibid., p. 103 n. 28. 119 Barnett 1928, p. 706. 120 Gonda 1963, pp. 91, 199. 121 Kuiper 1996 Middle Vedic literature. The gandharva found or proclaimed the "immortal names" (RV 10.123.4; 139.6); proclaimed (sacred) Speech in the womb (10.177.2); Viśvāvasu is asked to proclaim hidden things, "that which is reality and which we do not know" (10.139.5 123 ); the gandharva, "knower of the deathless" (amr̥tasya vidvā́n), should proclaim the supreme, hidden order (dhā́ma; AV [Śaunaka and Paippalāda] 2.1.2; RVKh 4.10.2). 124 I believe this knowledge of divine things, as well as the association with soma, are to be explained as due to the gandharva's dwelling in the highest heaven. Its abode, the heavenly waters, in which soma is found (before its descent to earth), are, as Lüders has shown, the highest sphere of the Vedic cosmos; in their affinity we find the vault or firmament (nā́ka-), the "back" or ridge (sā́nu-, pṛṣṭhá-) of the universe. Here is "the heavenly gandharva of the waters, watching over mankind" (nr̥ cákṣasa-; 9.86.36), standing on or above (adhi) the firmament (RV 9.85.12; 10.123.5) and "watching all his forms (i.e., beings)" (víśvā rūpā́ praticákṣāṇo asya; 9.85.12b). 125 The epithets indicate an omniscience of sorts; as to the latter one, we also find that "all forms" (víśvā rūpā́ in early Vedic frequently denoting all beings, or the entire world) are "his" (asya), implying a power over all creation 126 -cf. the lavish praise of Viśvāvasu in AV 2.2 as "the heavenly gandharva, sole lord of the entire universe". 127 Having his abode on top of the firmament, the gandharva occupies the function of transferring things -knowledge; the soma -from the celestial spheres to earth. 128 In Middle Vedic texts, the knowledge possessed by the gandharvas is, of course, ritualistic. We have tales of gandharvas interfering in ritual or liturgical matters, correcting the performers on various points (MS 1.4.12; JB 2.126; ŚB 11.2.3.7, 11.5.1.14ff). The tale of the 122 Oberlies 2005, pp. 104-105. 123 Transl. following Kuiper (1996, p. 241): yád vā ghā satyám utá yán ná vidmá. 124 The "gandharvic path of Order" (gā́ndharvīm pathyā̀m r̥ tásya) that Agni knows, according to RV 10.80.6, is interpreted by Geldner, and Lüders (p. 540 n. 2), as the correctly performed sacred formula, connected with the gandharva in 10.123.4 and 10.139.6. 125 Prati-cakṣ-can mean "behold" as well as "display, make visible" (cf. examples in Böhtlingk and Roth s.v.; Grassmann s.v.); Geldner takes the word here in the latter sense ("seine Farben alle offenbarend"; cf. also Haas,p. 214), while Brereton and Jamison translate, "gazing upon all his forms". "Watching all forms" is, however, a figure used elsewhere in connection with celestial beings; thus RV 10.139.3 (víśvā rūpā́bhí caṣṭe; of the sungod) or 10.136.4, where the "long-haired" shaman, leaving his body, "flies through the air, beholding all forms" (antárikṣeṇa patati víśvā rūpā́vacā́kaśat). 126 It may be noted that the all-god Rohita, the "ruddy" sun, is praised in almost identical terms in AV 13.1.11ab: "Rohita stood upright upon the firmament, bringing forth all forms, the young sage" (ūrdhvó róhito ádhi nā́ke asthād víśvā rūpā́ṇi janáyan yúvā kavíḥ). 127 V. 1: divyó gandharvó bhúvanasya yás pátir éka. Also RV 9.86.36, where the gandharva "rules over the entire world" (víśvasya bhúvanasya rājáse); note, however, that the "gandharva" here is Soma. 128 Kuiper (1996, pp. 249ff) has questioned the original celestial nature of the gandharva, arguing that it originates with the gandharva's occasional identification with Soma, or from a priestly "tendency to situate figures and events in heaven(s)" (p. 251). Instead, he attempts to locate this kind of being within his own theoretical framework, which posits a primeval war between gods and asuras and a resulting formation of the ordered cosmos by the victorious gods. Supporting himself on RV 10.10.4 (where he interprets the gandharva of the waters and the water-maiden as the ultimate ancestors of gods and humans), Kuiper sees the gandharva as a primeval being existing before the cosmogonic battle, and ascribes its knowledge of things hidden (pp. 239ff, 253; cf. Kuiper 1979, pp. 94-5) to this fact; thus the three-headed gandharva of the (much younger) JB already knows the outcome of the war. I see very little support for the interpretation of the gandharva's abode as being the primeval waters from which the earth arose; his watery abode is frequently located in heaven. As Oberlies has pointed out, the notion of the gandharva as guardian of the heavenly soma -which, probably already in Indo-Iranian times, was located in the celestial sea -is well in keeping with its celestial nature (Oberlies 2005; 2012, pp. 76ff; cf. 33ff, 79ff). gandharva Ūrṇāyu (PB 12.11.10; JB 3.76-7 129 ) is of great mythological interest here. It explains the origin of the aurṇāyava liturgy, on which JB has the following to say: Ūrṇāyu the gandharva was lusting for apsarases. He saw this sāman (liturgy). He praised with it. Whomever he desired he approached and won over with it, [saying,] "This one!" and touching her. Thus, this is a wishfulfilling sāman; this wish that he had, that wish came true for him. Whatever one may wish for, who praises with this sāman, that wish comes true for him. As Ūrṇāyu the gandharva saw it, therefore it is called aurṇāyava. 130 Then the story turns to a priest of the Aṅgiras clan, named Kalyāṇa (PB) or Śvitra (JB). The Aṅgirases were performing a sattra sacrifice for attaining heaven, but without result; searching for a solution, this priest came upon Ūrṇāyu, "who was swinging amidst the apsarases"; 131 as soon as the gandharva pointed to one of these, she became his. Ūrṇāyu taught him the wish-fulfilling liturgy, but had him promise not to claim himself as its discoverer (i.e., the one who had "seen" it through revelation). Kalyāṇa/Śvitra returned to the Aṅgirases, and with the help of the new liturgy their sacrifice became successful; but when asked about its origin, Kalyāṇa/Śvitra declared himself to have discovered it, and so he was left behind when the others attained heaven.
We also hear of knowledge of future events. In one myth on the frequent theme of the war between gods and asuras, the former seek information from a three-headed (triśīrṣan-) 132 gandharva who knows the outcome of the war, and what could change it (JB 1.125-27; variant in Baudhāyana Śrautasūtra 18.46 133 ). By seducing the gandharva's wife, and then eavesdropping on their conversation, Indra obtained the desired information.

Possessing women
Mostly, however, the gandharvas transmit their esoteric knowledge through the mouth of a mortal -a person possessed by one of them. Thus JB 2.126, 134 which tells of the wife of the brahmin Udara Śāṇḍilya, who was possessed by a gandharva (gandharviṇī, lit. "having a gandharva"). Through the wife, this gandharva warned the brahmin that the ekatrika sacrifice, which he intended to perform, was of a dangerous (dāruṇa-) kind; thus greatly surprising the brahmin, who had told no one about the sacrifice. In AiB 5.29.2 and KB 2.8.13, we find a "maiden possessed by a gandharva" (kumārī gandharvagr̥ hītā) quoted among the authorities invoked in a doctrinal dispute (on whether the agnihotra-rite is to be performed before or after sunrise). The most famous cases of gandharva-possession are found in the BĀU, 3.3.1; 7.1: during a brahmodya-, or contest in metaphysical knowledge, Yājñavalkya is on two occasions confronted by brahmins who have studied the Vedas in the house of Patañcala Kāpya. This 129 Cf. Caland 1970, pp. 237-38;1931, pp. 298-99. 130  man's wife and daughter were both possessed by gandharvas, who, through them, revealed knowledge on certain esoteric matters. Yājñavalkya is now questioned on whether he, too, possesses this knowledge (which he does).
What is considered spirit possession in pre-modern societies can, of course, often be identified as (especially mental) illness. The gandharvas and apsarases are indeed associated with madness, and so already in the AV. "The gandharvas and apsarases madden him who is about to go mad", states the TS. 135 In AVP 1.29 the apsarases are unmādayiṣṇavaḥ, causing madness; they are "mind-bewildering" (manomúhaḥ; AV(Ś) 2.2.5); they are also said to be fond of dice, and to preside over gambling (2.2.5; 4.38; 6.118; 7.109). "As gambling is repeatedly called an addiction, there seems to be a relationship with their causing madness"; 136 and here one might also point to the well-known Nala episode of the Mbh, where an obsessive gambling addiction is caused by possession by the evil spirit Kali. In AV(Ś) 6.111.4, the apsarases, together with Indra and Bhaga, are called upon to cure a person from insanity (lit. "give back" his mind); 137 in RVKh 4.8.3 we find the prayer (repeated in some younger texts 138 ), "The intelligence that is with the apsarases, the mind that is with the gandharvas -that intelligence which is divine or human; may it enter me now!" 139 These lines are part of a longer prayer or spell for intelligence or wisdom (medhā-); apparently, the gandharvas and apsarases had the power to bestow as well as take away one's reason. 140 It may be mentioned that the apsarases and their association with madness have often been compared to the nymphs of ancient Greece, 141 who were sometimes said to cause madness by their mere appearance, but could also, in their more positive aspect, possess a person who then -as a nympholêptos -became endowed with great intellectual skill or inspiration, and the gift of prophecy. 142 The question of a common origin of these beliefs is not easily answered; while there certainly do exist similarities between the apsarases and the nymphssuch as otherworldly beauty, and being inhabitants of trees and rivers -similar beings can be found also in the myths and folklore of non-Indo-European peoples. The association with madness, however, is certainly intriguing.
There also seems to be a more positive connection between gandharvas and apsarases, and mind or intellect. 143 As we have seen, they are implored to give (back) a person's manas, and 135 TS 3.4.8.4: gandharvāpsaráso vā́ etám ún mādayanti yá unmā́dyaty. 136 Oberlies 2009. In AV 6.130.1, the apsarases are implored to induce lovesickness (smará-) in a person; another kind of loss of mind, as it were. 137 "This fits well with the notion that madness was considered to be characterized by the mind leaving the body; and in order to become sane, it must be returned." (Zysk 2009, p. 188.) Cf. also Baudhāyana Śrautasūtra's (18.4.396;text and German transl. in Gotō 2000, pp. 100ff) version of the legend of king Purūravas and the apsaras Urvaśī, where the love-smitten apsaras stops the king's chariot by making an illusory hole appear in the road, then making it disappear, leaving the king wondering whether he is going mad (dr̥ pya-). 138 I refer to Bloomfield 1906, p. 343. The verses are later used for invoking wisdom to enter a Veda-student; cf. Gonda 1975, p. 168. 139 yā medhā apsarassu gandharveṣu ca yan manaḥ/daivī yā mānuṣī medhā sā mām āviśatād iha. 140 Elsewhere in this prayer, this wisdom is called gandharvajuṣṭām, "enjoyed by the gandharvas". While several other divinities are invoked to bestow it, it is, notably, said to actually be among the gandharvas and apsarasesnot with any other deity. This indicates, I think, that these beings could have possession of a person's mind and intelligence, and seize or withhold it at will. 141 Cf. e.g. West 2007, pp. 284-92;Oberlies 2009. 142 Cf. Connor 1988more generally, e.g., West 2007, p. 287. 143 Cf. Wijesekera 1994a bestow medhā upon him. In TS 1.7.7.1 and VS 30.1 (quoted in ŚB 5.1.1.16), we find the line, "May the heavenly gandharva, purifier of thoughts, purify our thoughts". 144 In RV 10.11.2a-b, "the gandharvī, the water-maid" (gandharvīŕ ápyā ca yóṣaṇā), is invoked by the poet to "protect my mind" (pári pātu me mánaḥ); in 10.139.5, Viśvāvasu, the "heavenly gandharva" (divyó gandharvó; obviously identical with the one of TS and ŚB), is invoked for similar purposes: "to stimulate our thoughts, to aid our thoughts" (dhíyo hinvānó dhíya ín no avyāḥ).
As appears from the cases cited earlier, the gandharvas possess -as it seems, exclusivelywomen. 145 This brings us to the subject of spirit-possession studies in comparative anthropology and religion. It is a well-known fact, albeit variously interpreted, that women are particularly (though by no means exclusively) prone to possession by spirits or deities; this is, or has been, the case in culturally unrelated societies around the world. I. M. Lewis' study of possession cults, 146 which is probably the most widely read work to date on the subject, argued that spirit possession is most prominent in marginalized groups (among which women are usually one); individuals belonging to these groups may, according to Lewis, raise their status and make their voices heard by letting themselves become the instruments of spirits or deities, who are believed to speak through them. As such, possessed persons often become the center of local cults, and their words carry great authority. While Lewis' theory is still widely influential, there are, of course, others; not least the high occurrence of possession among women has been the subject of a fair amount of theory-construction in a variety of fields (psychology, gender studies etc.). 147 Thus, traditional gender roles, which make women the "passive", "receiving" sex, have been invoked as one important reason why women are entered by spirits and become their passive instruments. Then there is the sexual aspect: the fact that the possessing spirits and deities are mostly of the male gender, and the possessed females are thought of as objects of their enjoyment. 148 Both these kinds of approach -the socio-anthropological and the psychological -seem to be applicable to the Vedic situation, as far as we can glimpse the beliefs and realities behind the legends. The wife and daughter of Patañcala Kāpya in the BĀU are clearly in the center of a kind of possession cult, filling the function of oracles and being questioned on esoteric matters. 149 Whether these cases are actual historical reality or not is unimportant; what matters here is that such cults obviously were in existence. An actual case of spirit possession 144 divyo gandharvaḥ ketapūḥ ketaṃ naḥ punātu. 145 Though the apsarases are frequently said to cause insanity, I cannot find any explicit reference to possession by them. Nonetheless, possession was most likely thought to be the means by which they induced madness in people. If so, the question arises as to whether they only possessed men, like the gandharvas possessed women. As the gambler or kitavá-apparently was one of their main targets, this seems to have been the case; for the kitavá-as typically a young man, see Falk 1986, p. 99. The gambling hall or sabhā-is well known to have been an exclusively male area. 146 Lewis 1975. 147 See, for example, Sered 1994, Keller 2002, Schmidt and Huskinson 2010 theory still remains one of the most valuable, despite the criticisms levelled in some of these works. 148 For the erotic undertones of possession, and the belief that spirits mainly possess persons of the opposite gender, cf. Lewis, pp. 58ff, 84. 149 Cf. Smith 2009, p. 230: "Did Bhujyu Lāhyāyani simply wander into his friend Patañcala Kāpya's house on a couple of occasions and discover his wife and daughter occupied with housework, possessed by gandharvas, ready to take questions? This is highly unlikely. More likely, given the evidence of later texts, a ritual was taking place and Bhujyu Lāhyāyani showed up for the occasion." appears, on the other hand, to be referred to in the AiB and KB, where a certain "maiden possessed by a gandharva" is cited as an authority on matters of doctrinal dispute (even though the view of the maiden/the gandharva on these matters is ultimately rejected by the authors). It seems clear that possession by gandharvas made it possible for certain women to take part in learned discussions from which they were otherwise barred; such as brahmanical disputes on the subject of ritual.
At the same time, it can't be denied that there is a sexual dimension to the possession by gandharvas -beings who are so often referred to as "fond of women" (strīkāma-). We have already quoted AV 8.6.19, where mention is made of the "women-sharing (strībhāga-) gandharvas" who cause the embryo in the womb to die; this refers, most likely, to a fatal sexual enjoyment of pregnant women. 150 Very likely, this enjoyment was thought to take place through possession; we have already referred to the fact that demon possession was believed, in Vedic as well as younger times, to be the cause of miscarriage. Indeed, there seems -judging from the texts -to be no better explanation as to why gandharvas only possess women. (Only in much later texts, like the medical compendia of Suśruta and Caraka, do we meet with possession by gandharvas also of men or boys. 151 ) Spirit possession never became an integrated part of the brahmanical religion, nor in the Buddhist or Jainist traditions; possession-cults were mainly confined to the more popular religious life, as a consequence of which possession has long been a neglected subject in the historical study of Indian religions. Some of the lacuna has recently been filled by Frederick M. Smith's extensive survey of possession in South Asian religions, from Vedic to modern times. 152 Dealing with questions of comparative anthropology, Smith notes that possession has occurred especially in women and children; this holds for almost all the epochs treated, from the gandharva possession of Vedic times to classical "Hindu" beliefs and even up to the present day. 153 Of interest here is also his survey of the terminology of ancient Indian spirit possession, where two terms in particular are singled out as the most prominent ones: formations from the verbal root viś-"to enter" (with prefixes pra-or ā-), and the verbal root gr̥ h-"to grasp, to seize"; the former occurs in connection with more "positive", often ritually controlled forms of possession, in which a deity "enters" a human being (though pra-viśusually refers to possession independent of the possessed person's will), while cases where gr̥ h-is used are of a more sinister nature: the possession is here caused by spirits or demons of disease or madness, who forcefully take control over ("seize") a person's body. From this root is derived the word graha-, referring to demons of sickness (several times already in AV). 154 It is formations from gr̥ h-that are used in depictions of gandharva possession; thus the 150 Kuiper (1996, p. 245) points out that the designation as strīkāma-seems to occur only in the context of the myth of the soma-barter. While this may be true, the lustful nature of these beings is well in keeping with atharvavedic and later materials. 151 Suśruta Saṃhitā, 6.60.10, declares the symptoms of gandharva-possession to be laughing and dancing, and a fondness of singing, perfumes and garlands; thus, "in gewisser Übereinstimmung mit dem Grundcharakter der Gandharven", as Hillebrandt (1987, p. 183) noted. In Caraka Saṃhitā, 6.9.21.4, it is rather the gandharva-like character of a person that causes the possession: "The gandharvas attack a person of pure behavior who is fond of hymns of praise, singing, and musical instruments, who is fond of other men's wives, perfumes, and garlands, generally on the twelfth and fourteenth lunar days" (quoted from Smith 2009, p. 409). 152 Smith 2009. 153 On spirit possession and women, see esp. pp. [68][69][70][71][72][73][74][75]430ff,545ff;on gandharva possession,Index,s.v. graha,grā́hī. gandharvagr̥ hītā-("seized by gandharvas") women in AiB, KB,and BĀU. 155 This confirms the general impression from the source-texts that possession caused by gandharvas was an unwelcome thing -usually leading to insanity -though it could be used to the advantage of the husbands or relatives of the possessed woman (divination).
Such divination rituals are well attested in somewhat later times. Of special interest in this connection is Smith's study 156 of a medieval tantric ritual of oracular possession, called svasthāveśa-, or simply praśna-, "questioning" (prakrit pasiṇa-; later, falsely re-sanskritized as prasenā-). In this ritual, a spirit (itself often called prasenā) is invited to enter an object, a body-part, or a person -usually a young girl or a child; the spirit is then questioned on future events. In one form of the ritual, a maiden serves as an oracle, while being in a state of trance and, sometimes, looking into a mirror. 157 According to Hemacandra, 158 a deity (devatā) is made to enter the mirror, through which it reveals the desired knowledge to a young girl (kanyā) looking into the mirror. The question answered through this divination practice is here, specifically, said to concern the time of one's death. Now, while this and the other texts collected by Smith all belong to medieval times -the oldest being from the second half of the first millennium C.E.it may be noted that the Sāmavidhāna Brāhmaṇa, 3.8.1ff (one of the very youngest Brāhmaṇa texts, but certainly older than any of the tantric texts treated by Smith), prescribes a very similar ritual, with the purpose of preventing one's rebirth after death. Here, a kanyā, with plaited hair (śikhaṇḍinī-) and noose in hand, serves as oracle in a preparatory, nightly ritual; she reveals to the person wishing not to be reborn the year, halfyear, season, month, etc., down to the day or night and muhūrta when he is going to die, thereupon to be reborn (3.8.3). With this knowledge, that person may then, using spells and penance, avert the rebirth awaiting him, and instead attain the realm of air (ākāśa-) after death. The Dīgha Nikāya, 1.26, in a list of divinatory and magical practices prohibited for Buddhist monks, mentions kumārīpañha-, "questioning a young girl"; the word appears between ādāsapañha-, "questioning a mirror", and devapañha-, "questioning a god", and obviously refers, as Rhys Davids (following Buddhaghosa) translated it, to "Obtaining oracular answers from a girl possessed". 159 In devapañha-too, a girl serves as the oracle, according to Buddhaghosa's 5 th century commentary; but here, it is a temple prostitute (devadāsī), while the kumārī is said to be of good family and respectable. Using young girlsusually in a state of trance, but perhaps also insane ones -as oracles was, as it seems, a longstanding tradition in ancient India. 160 Now, it may be suggested -but this is only a speculation -that the association with possession could explain some of the more prominent traits of gandharvas; such as their 155 The word gr̥ hīta-is, however, not used in the story of Yavakrī and the apsaras (pace Smith, p. 228); nor is any other word from the same root. 156 Smith,chap. 11. 157 "One might argue that this is not possession of the girl; rather, it is an allied divinatory practice. However, in South Asia, people, especially women, are considered possessed if they transmit such messages in trance states." Ibid, p. 431. 158 Yoga Śāstra 5.173-6; quoted in Smith,p. 431. 159 Cf. also the entry under this word in Rhys Stede 1952 [1921]. 160 Cf. possibly the female vipraśnikās consulted by king Dhr̥ tarāṣṭra Vaicitravīrya (Kāṭhaka Saṃhitā 10.6) to find out the cause of a disaster that has befallen his domain (they reveal that it is caused by the black magic of an insulted brahmin). While we learn next to nothing from the text about these vipraśnikās, it is remarkable that women seem to have been employed as psychics at the, otherwise male-dominated, Vedic royal court. fondness of garlands, ornaments, scents, music and singing. Lewis gives numerous examples of possession cults centered around women, where "the spirit" speaking through the possessed woman demands luxuries such as clothes and ornaments; song and music are frequently part of the "treatment" of the possession. The often very specific requests of the spirits are believed to mirror the particular nature of these beings; the Islamic jinns, for instance, were in Somalia "thought to be consumed by envy and greed, and to hunger especially after dainty foods, luxurious clothing, jewellery, perfume, and other finery". 161 In the medical compendia of Caraka and Suśruta, people possessed by gandharvas are indeed said to be hankering for the same things as these beings: garlands, scents etc. 162 The youthful and careless nature of these deities, to which we will return, is, however, also likely to have contributed to these conceptions.

Apsarases and men: fleeting relationships and gandharvic jealousy
The conception of "women-desiring" gandharvas seizing females and, at will, revealing sacred knowledge through them, brings us to what may be the core of the Vedic mythology around gandharvas: there takes place a kind of exchange during the possession, wherein the gandharvas enjoy mortal women who are often married, but, on the other hand, may provide their husbands or relatives with esoteric knowledge. This is, more or less, the same kind of exchange as that which takes place in the myth of the bartering of soma: there, a sacred and powerful substance is traded by the gandharvas for Vāc -who, sure enough, embodies the Vedic sacred formulas, but is desired by the strīkāma-gandharvas solely as an object of sexual enjoyment. For the somewhat more pious gods, however, her presence is of the highest importance, and so they have to win her back. The case is similar in the story of Ūrṇāyu, who, knowing a wish-fulfilling liturgy capable of ensuring mortals of heaven, himself put it to no better use than for obtaining women. The gandharva was, as we have seen, thought to dwell in the highest celestial spheres, where he was in direct contact with sacred substances such as soma, and in possession of superhuman knowledge; this heavenly realm was, however, thought of merely as a depository, from which these things were seized by the gods in ancient times (e.g., in the various myths of the soma-robbing, finding the hidden Agni etc.), and from which they may be brought to earth by the gandharvas, who, as intermediaries between heaven and earth (Oberlies), bring the sap to the soma plant, embryos to mortal wombs, and sacred knowledge to certain humans. The gandharvas themselves, being merely the keepers and carriers of these things from the "depository" realm, seem to make no actual use of them -except, sometimes, for the sake of enjoying women. A rather clear example is the primeval war between gods and asuras, as told in some versions in the JB. 163 Here, it is related that "the Kali gandharvas" 164 did not take part in this war, maintaining a neutral stand (antasthā-); yet 161 Lewis p. 75. 162 See n. 148 above. Similarly in Viṣṇudharmottara Purāṇa 1.231.36ff, quoted by Thite (1987), p. 58. Noting that this passage prescribes music as part of the possessed person's treatment, and that similar "cures" for possession are to be found in other cultures, Thite suggests a connection between the gandharva's association with music, and his possessing people. Smith (p. 230) proposes to see gandharva-possession in "a context … in which music was played as part of a ritual to abet the onset of trance states, such as possession", but provides no further basis for this assumption. 163 Cf. translations in O'Flaherty 1985, pp. 86ff;Bodewitz 1990, pp. 71-2, 86-7. 164 Also mentioned in AV 10.10.13; what makes them distinct from other gandharvas is not clear. after the war asked the victorious gods for a share in the conquered worlds, as they had been supporting their side in their minds (1.154-55). Then there is the story of the three-headed gandharva who knew the outcome of the ongoing war, and how to change it, yet did not enclose this information to the gods (nor to the asuras), who had to obtain it through trickery (1.125-27). The knowledge or sacred substances possessed by the gandharvas are, so to speak, not "activated" or used to their full potential, until they leave these deities.
Interestingly, a woman appears in an intermediary function also in the tale of the threeheaded gandharva: the gandharva's wife, who is seduced by Indra in a successful attempt to obtain the desired information; here it is, thus, the gandharva who is cuckolded, not the other way around. This tale throws some further interesting light on the gandharvas' relationships with women. Of the three-headed gandharva it is said that "he had a boat-mansion 165 floating about in the waters". 166 "The waters" are apparently the heavenly ocean, the traditional home of the gandharvas; āpaḥ or its locative form, apsu, being often used, without further qualification, to denote the celestial waters. 167 Now, a reason is given for the gandharva's unusual choice of abode: although the readings of the manuscripts are corrupt, Hoffmann 168 is no doubt right in his emendation to sa herṣyur āsa -"he was jealous-natured" (1.125). Hoffmann supports this reading on a parallel in JB 3.197, where it is said of the demon 169 Asita Dhāmnya that he protected his daughter's virginity in a similar way: "Now, Asita Dhāmnya was jealous (īrṣyur āsa). He had a palace in the air." 170 (As in the case of the gandharva's wife, these efforts at protection are ultimately frustrated.) That the palace is said to be in the sky would seem to support our assumption that the gandharva's "boat-mansion" is floating in the heavenly waters; one is reminded of the aerial "cities of the gandharvas" (gandharva-pura-or -nagara-), a kind of fata morgana often mentioned in later literature. 171 The gandharva of the tale thus has a jealous nature, and seeks to guard his wife at any costs. That this is not an individual characteristic of this specific gandharva may be established 165 The version in Baudhāyana Śrautasūtra 18.46 has a "golden boat" (hiraṇmayyā nāvayā; O'Flaherty's rendering "a golden palace" must be a slip). 166 tasya hāpsv antar naunagaraṃ pariplavam āsa. This is one of the earliest occurrences of the word nagara-, later meaning "city", though cities may still have been unknown at the time of the text's composition; still, the compound nau-nagara-has been rendered as "ship-town" (Bodewitz), "boat-city" (O'Flaherty 1985), "Schiffsburg" (Hoffmann 1960, p. 7). Kuiper (1996, p. 238), however, gives "house-boat", commenting: "In Dravidian, from which nagaram has doubtless been borrowed … nakar 'town, city' originally denoted a single building: a temple (Tamil), a palace (Old Tamil, also Telugu nagaru) and even a 'house, abode, mansion' (Old Tamil, in Akanāṉūṟu 15) … The Old Tamil Sangam literature dates from the first centuries A. D., and naunagaram, lit. a 'house on a boat', reflects the older meaning of nakar in Dravidian." Personally, I suspect that nagara-in the text concerned denotes something bigger -a mansion or a palace -as the proper name Nagarin occurring in some Brāhmaṇas would seem rather pointless if simply denoting a "house-dweller." 167 Cf. e.g. JB 1.292: "the lightning in the waters" (vidyud apsu); JUB 1.34.4 (the sun and moon being seen in the waters, apsu); ŚB 7.5.1.8 (the sun giving heat deep in the waters). 168 Hoffmann 1960, p. 7. 169 An "Asita Dhānva" is mentioned as the leader of asuras in ŚB 13.4.3.11; one line in JB 3.197 implies that his kin are rakṣases. 170 atha asito dhāmnya īrṣyur āsa. tasya hāntarikṣe prāsāda āsa. For the entire story, see Caland 1970, pp. 269-70;O'Flaherty 1985, p. 95. 171 See references in Hopkins 1915, p. 157;Böhtlingk & Roth 1855-75, s.v. gandharvanagara, gandharvapura. The term does not appear in Vedic, though the Ṣaḍviṃśa Brāhmaṇa, 6.8.13, mentions a "palace in the sky" (ākāśe rājakulaṃ) in a list of bad omens; Sāyaṇa glosses this as gandharvanagara-. Might the "golden palace" (hiraṇyavimitāni) of the gandharvas and apsarases, appearing before Purūravas in ŚB 11.5.1.11, be related to the later conception? through a comparison with JB 2.269-72; the story of the brahmin Yavakrī. 172 This man used to take advantage of his brahmanical status and power for having whatever woman he wanted; "Whomever he called upon would make love to him and then die; and who did not make love to him would also die". 173 (This easily reminds one of Ūrṇāyu's ability, though in a more sinister form.) One day he called in this way upon the wife of the brahmin Yajñavacas Rājastambāyana. When her husband later found her ornamented and crying, prepared to make love to Yavakrī and then die, he performed a fire sacrifice; and from the oblation rose an apsaras in the likeness of the wife. She was sent to Yavakrī in the wife's place. As they were about to have intercourse, the apsaras started giggling and showed the brahmin the hairy soles of her feet (lomaśau … adhastāt pādāv), thus revealing her true nature (cf. the "hairy boy" of AV 4.37.11, donning a pleasant guise before women 174 ); meanwhile, Yajñavacas had been performing another offering, this time producing "a jealous (īrṣyu-) gandharva with an iron axe in his hand". 175 "He [Yajñavacas] said to him, 'That wife of yours has gone to Yavakrī!'"; 176 and so the gandharva went there, finding the brahmin in bed with the apsaras. Terrified, Yavakrī asked him about some penance (prāyaścitta-) to atone for his offense; the gandharva told him to cut the heads off all his and his father's cattle before sunrise. Yavakrī did not survive the night; he was, according to the text, either killed by the gandharva, or by a local carpenter who wanted to put an end to the outrageous cattle-slaughter. There are a few hints in the text that seem to imply that Yavakrī was in fact possessed, or perhaps was so in an earlier version; people seeing him behead his cattle are said to have told each other, "Yavakrī has gone insane!" (adr̥ pad yavakrīr) -gandharvas and apsarases were, of course, believed to be responsible for mental illness in people -and his father then replied that his son seemed to be "driven by gods" (deveṣito). 177 What is important for the present discussion is the fact that the gandharva is described as īrṣyu-, jealous, and that it is jealousy that drives him to take Yavakrī's life; apparently, Yajñavacas could not simply produce a gandharva and command him to kill Yavakrī, but had first to make his enemy cohabitate with an apsaras, thereby provoking the gandharva's wrath.
The motif of the gandharvas' jealousy is also present in the famous legend of Purūravas and the apsaras Urvaśī, as told in the ŚB (11.5.1). This is, basically, the traditional tale of a mortal man espousing a supernatural woman, only to lose her after breaking some taboo; 178 but it 172 Cf. Caland 1970, pp. 190-94;O'Flaherty 1985, pp. 105-11. (The version edited and translated by Caland is somewhat shorter than the one in Vira and Chandra's edition, and leaves out some of the passages discussed here.) 173 sa ha sma yām acchābrūte yā ha smainaṃ kāmayate mriyate ha sma, yo ha smainaṃ na kāmayate mriyata u eva. (2.269.) 174 Though references to an ugly, "true" form of the (usually handsome) gandharvas and apsarases are rare, enticing and seductive beings of folklore often are, partly or at some times, ugly or hairy; even when appearing in a beautiful guise, lower parts of the body (which are hidden under clothes) are said to be hairy. Thus the forest nymphs of Scandinavian and German beliefs is depicted as a beautiful woman bent on seducing wanderers, but can be recognized by her having furry legs, a tail, or goat's feet (Mannhardt 1875, p. 95 n. 1, 128ff). In a wellknown Arabian and Jewish legend, the beautiful queen of Sheba, being the daughter of a jinni and a mortal man, was recognized as such by king Solomon, who tricked her into lifting her skirts and revealing her hairy feet. 175 hāyaḥkūṭahastaṃ gandharvam īrṣyum (2.270). What follows is part of the motif of "the shattered head" (Witzel 1987), wherein an offender is threatened with having his head split open by a supernatural being. 176 [a]sau te jāyā yavakrīyam abhyagāt. 177 The very same word is used for the possessed muni-of RV 10.136, for which see more below. 178 Cf. Aarne-Thompson No. 400: "The Man on a Quest for his Lost Wife." also appears to reveal much regarding the nature of the gandharvas and apsarases. The "taboo" in this story is laid down by the apsaras when she agrees to live with Purūravas: if he would ever show himself naked to her, she would leave him. 179 When Urvaśī had lived with Purūravas for a while, and had become pregnant, "the Gandharvas said to one another, 'For a long time, indeed, has this Urvaśī dwelt among men: devise ye some means how she may come back to us.'" 180 In the night, they robbed away a sheep that was tied to Urvaśī's bedstead; the apsaras called for Purūravas to go after the thieves, and as he hurried out in the night without putting on any clothes, the gandharvas caused a lightning-flash to light up the place and reveal the undressed Purūravas to his wife. (This event is already alluded to in v. 3 of the dialogue between Purūravas and Urvaśī in RV 10.95.) The apsaras consequently disappeared. Purūravas, having "wandered all over Kurukṣetra" in his sorrow, eventually came upon his lost wife as she and other apsarases were sporting in a pond in the shapes of water-birds; after several attempts at persuading her to return to him (the dialogue is derived from RV 10.95), he is told to return to the place after one year has lapsed. When he returns, a golden palace (hiraṇyavimitāni) has appeared on the spot; Purūravas is there met by Urvaśī, who declares, 'To-morrow morning the Gandharvas will grant thee a boon, and thou must make thy choice.' He said, 'Choose thou for me!' She replied, 'Say, Let me be one of yourselves!' In the morning the Gandharvas granted him a boon; and he said, 'Let me be one of yourselves!' 181 (11.5.1.12,transl. Eggeling.) The gandharvas then instruct Purūravas on how to perform a fire-sacrifice, which makes him one of them and allows him to reunite with his wife.
The gandharvas here appear as a group that does not want their women to establish longterm relationships with outsiders, or at least not with mortals. 182 (Though preventing such relationships must have been quite a task, considering the well-known licentiousness of the apsarases.) They break up Purūravas and Urvaśī's relationship because they think she has dwelt too long among mortals; when the apsaras finally gives in to Purūravas' pleadings and agrees to take him back, he is told that he first has to become one of the gandharvas. Even if the story's happy ending, which is not alluded to in RV 10.95, should be considered a late, brahmanical addition reflecting the priestly sacrificial ideology, 183 it nonetheless appears to tell us something about the conception of the gandharvas: though both they and (at least in post-Vedic mythology) the apsarases frequently indulge in fleeting relationships with humans, they seem to marry exclusively within their own group. This is in keeping with the epic legends which have apsarases consorting for a while with mortal kings and heroes, only to later abandon them and the children that usually are the outcome of the relationship. Only in paradise, it seems, are the pious deceased -and especially heroes slain in battle -actually 179 "Originally", it may have been the apsaras who was not to be seen in her true form, as comparative folklore (e.g., the story of Cupid and Psyche) suggests; thus Oldenberg 1894, p. 253, Keith 2007. 181 gandharvā vai te prātar varaṃ dātāras taṃ vṛṇāsā iti taṃ vai me tvam eva vṛṇīṣveti yuṣmākam evaiko 'sānīti brūtād iti tasmai ha prātar gandharvā varaṃ daduḥ sa hovāca yuṣmākam evaiko 'sānīti. 182 But apparently not with several gandharvas, as their relationships are usually promiscuous; see below. 183 Cf. Oldenberg 1894, p. 254 n.1;Keith 2007, p. 183; for a different opinion : Geldner in Pischel and Geldner 1889, p. 259. married to apsarases, 184 though this theme, too, may be a late development; it appears first in late Vedic texts. 185 The case of Purūravas is similar: he is not accepted as a proper husband of Urvaśī until he himself has become a gandharva and joined their numbers in heaven.
A late expression of the belief in the gandharvas' dangerous jealousy is found in the Virāṭaparvan -the fourth book of the Mahābhārata. 186 As the five Pāṇḍava heroes and their wife Draupadī, during their last year in exile, live disguised as servants at the court of king Virāṭa, Draupadī -who has donned the guise of a maidservant (? sairaṃdhrī-) and, pretending to be unmarried, worries about her beauty attracting men -seeks to scare away suitors by claiming to be married to five bellicose gandharvas. These supernatural husbands will not tolerate any man making approaches to her: I am not to be obtained by Virāṭa, nor by any other man at all. I have for husbands five youthful gandharvas, O splendid one, sons of a certain glorious gandharva king. They always protect me; so I am dangerous to approach … The man who covets me like other, lowborn women, he will that very night enter another body [i.e., die]. 187 Later on, she is forcefully approached by the king's commander, Kīcaka, who doesn't heed her warnings about the wrath of the gandharva husbands. She eventually has one of her real husbands, Bhīma, secretly kill the harasser, squeezing him into a ball with his bare hands; Draupadī then declares to everyone that this was done by the jealous gandharvas. After Kīcaka's vengeful kinsmen are similarly slain by Bhīma for assaulting Draupadī, the king wants her to leave the court for everyone's safety, but has his queen telling her this, not daring to do so himself -as a man -for fear of the gandharvas (4.23.8-10).

Weddings and initiations: the gandharva and rites de passage
A similar struggle for women as seen in these tales may be pointed out in the myth of the bartering of the soma: the gandharvas ask the gods for Vāc -not in her function as goddess of sacred speech, but simply as an object of enjoyment -but are made to promise (in the ŚB version) not to restrain her against her will (as they may have been prone to do with their women; cf. the three-headed gandharva in JB). The competition for her ends with their defeat and Vāc's return to the gods. There is some evidence that connects this myth with the notion of the gandharvas' right to a girl before her marriage, and their unwillingness to let go of her even after the wedding. The late Vedic wedding-hymn recorded in the Kāṭhaka Gṛhyasūtra has some stanzas that are based on the myth, speaking of the goddess who is being evoked by two contending parties: We, both gods and gandharvas, call upon you together in contest. Go to those of them (us) whom you desire, Sarasvatī! 184 Cf. the materials in Hara 2001;cf. Hopkins, pp. 161, 163. Note that in Mbh 11.26.13 The gandharvas and apsarases are mentioned throughout the hymn; in the half-verse immediately preceding the verses quoted (v. 18c-d), the singer (that is, the bridegroom) refers to his own "gandharva with the head" -the male organ -which wins over the bride for him: "May the gandharva with the head make you stay with me." 189 The Vādhūla Śrautasūtra's account of the contest for Vāc follows, word for word, the version given in TS 6.1.6.6; the only difference being that it includes four stanzas from the above-mentioned wedding-hymn. This is the more interesting as the prose-text from TS makes explicit mention of marriage. I quote here the relevant part of the Vādhūla text: The gandharvas uttered a sacred formula (brahman); the gods sang. "Those people who existed formerly", was the sacred formula the gandharvas uttered; "Those who formerly brought weal to them -for them the gandharva with the head gave heat before the gods. Those people who existed formerly, before the former onesthe one with the head, the son of Subhrū (?), also gave heat for them before the sun." "The one first choice of women", was the song (gāthā-) that the gods sang, "in which is found this entire world -that song will I sing today, that which is the highest glory of women. Support this one, fortunate Sarasvatī, rich in rewards! We sing to you, prior to all existence!" She turned to the singing gods. Therefore women desire one who sings; women become desirous of him who knows thus. 190 The TS text concludes: "So if there is in a family one person who knows thus, men give their daughters in wedlock to that family, even if there be other (wooers) in plenty." 191 All these stanzas appear, though in a different order, in the hymn in KGS. While the Vādhūla text presents some of them as a (profane) song, gāthā, and the others as a bráhman or sacred formula, they are in fact all, as Caland noted,192 part of the same wedding-song. He also pointed out 193 that the association of the myth with weddings appears already in MS 3.7.3, which concludes its version of the story with the following words: Therefore a song is sung at a wedding. Therefore one who sings is dear to a woman. That is why one knowing thus, singing a song, marries [lit. "grasps the hand (of the bride)"]. Then the two (the married couple) age together. They live their whole lifespan.  192 Caland 1929;, p. 159. 193 Caland 1928, p. 158. Cf. Ludvik 1998 tásmād vivāhé gā́thā gīyate tásmād gā́yant striyā́ḥ priyás. tád yá eváṃ vidvā́n gā́thāṃ gā́yan hastáṃ gr̥ hṇā́ti sáṃ hí jīŕyataḥ sárvam ā́yur ito nā́rtiṃ nītas. (Transl. Ludvik, ibid.; brackets mine.) While this dimension of the myth is no doubt secondary, it is thus, nonetheless, very old. It appears from the wedding-hymn that the bride was thought of as the object of a struggle between two parties, here referred to as gods and gandharvas. We may point here again to the belief recorded in AV 14.2.9, that the gandharvas and apsarases, hiding in trees, sought to attack the wedding-procession. Vasilkov 195 has suggested that this notion may be connected with a custom known from several societies, in which, at a wedding, a mock-battle between two contending parties is fought over the bride. There would, in that case, have been people present who impersonated the gandharvas and apsarases. Vasilkov believes that these people are to be recognized in the "long-haired people" (keśíno jánā) and "young sisters" (jāmáyo … yuvatáyo), dancing in the bride's house and wailing over her departure, who are mentioned in the same hymn (14.2.59-61). These have some characteristics in common with the gandharvas and apsarases -youthfulness, dancing, long hair (for which see below). Who, then, are these people? Vasilkov's study suggests a connection between the gandharvas and apsarases of mythology and age-set groups of young boys and girls who have not yet entered the adult, married life. The bride would have belonged to such a group before her marriage; thus the notion that she used to be married to the gandharva(s). This assumption may find some support in the stanzas 14.2.59-61, which seem to imply that the bride used to be one among the "dancing" youths who are now wailing: If these long-haired people have danced together in your house, doing evil through wailing -may Agni and Savitṛ release you from that sin! If this daughter of yours has wailed with dishevelled hair in your house, doing evil through wailing (etc.).
If these young sisters have danced together in your house, doing evil through wailing (etc.). 196 The word jāmí-, "sister", used for the young women (yuvatáyo) in v. 61, is of interest; when Viśvāvasu is implored, in the same hymn (v. 33), to leave the bride, he is told to "seek out a sister (jāmí-) who, [though] mature, dwells in her father's house; that is your share by birthseek it out!" 197 But in the two following stanzas (34-5), he is instead implored to return to the apsarases, who are his "kin" (janítram; 35) and his "wives" (jāyā́; 36). The imprecation to leave the mortal woman to her husband and go to his own wives, the apsarases, occurs, as we have seen, elsewhere in AV. There thus seems to be a parallelism between the apsarases and the unmarried young women who are considered to belong to the gandharva(s). Vasilkov suggests that jāmí-was the designation for those unmarried girls who belonged to the youthsocieties.
There can be no doubt as to the fact that unmarried girls were considered "wives" of the gandharva(s); and, as we have seen, propitiatory offerings to Viśvāvasu or the gandharvas in general were performed either when the girl entered marriageable age, or before she was married off. The evidence for unmarried young men being connected to this kind of beings is sparser. Conceivably, the bamboo-staff which is handed over to the snātaka-, the graduated Veda-student, and which is addressed with the words, "Thou art the gandharva Viśvāvasu; protect thou me, guard thou me" 198 (Jaimini Gṛhyasūtra 1.19; transl. Caland), may once have had some function cognate to the girl's offerings to the gandharva; the student, freed from his vow of celibacy (brahmacarya-) and entering the life of a married householder (gṛhastha-), is in a position similar to the girl who is about to marry, finding himself between two major lifestages. The gandharva's role as a being of transfer, responsible also for safely establishing the bride in her new home, may be of importance here too.
I would suggest, with some caution, that the single gandharva of old Vedic times filled the function of a tutelary deity of young boys and girls who had not yet entered the adult, settled life. This single gandharva seems to have lived on in the -generally conservative -domestic ritual of the Gṛhyasūtras: it is to be noted that it is frequently the original gandharva, Viśvāvasu who appears in connection with the rites of passage preceding marriage or at the end of studentship; only occasionally is he given the qualifying epithet gandharvarāja-, as if to show why he, among all gandharvas, appears alone in this role.
Vasilkov has put forward the theory that the gandharvas and apsarases were seen as the celestial counterparts of the youths living together in a "men's house", known from many tribal societies, where they indulged in promiscuous relationships. The unmarried girls serving the young men as objects of their common enjoyment (an institution also known from ancient Iran) would correspond to the apsarases, who are indeed frequently described as celestial courtesans in the classical literature. My own interpretation, as given above, is slightly different, and attempts to pin down the "original" function of the single Vedic gandharva, as reflected not least in the later domestic ritual; but I think Vasilkov has collected enough evidence (to be discussed below) to establish a connection between the gandharvas and the "men's societies" of Vedic times. Conceivably, this is a later development; the single gandharva of the RV and parts of the AV is more likely to have been a tutelary deity of the pubescent youths, rather than identified with them. This is also the picture given by the domestic rites. The exact relationship between the gandharva(s) and the "men's societies" is, however, difficult to decide on due to the scarce information we have on this kind of institution in ancient India. What can be ascertained is, first of all, that a "men's house" can be identified in the oft-mentioned sabhā-or "assembly hall"; 199 second, that this sabhā is fairly clearly put in connection with gandharvas, as well as prostitutes.
Before proceeding with a survey of Vasilkov's evidence, some words should be said on the nature of the Vedic sabhā. Well known from Vedic and epic literature to have been an exclusively male assembly hall with various functions -one of the most prominent being as a gambling hall -we have some very clear statements as to what kind of person used to frequent the sabhā. According to ŚB 13.1.9.8 and Taittirīya Brāhmaṇa 3.8.13.1, the sabhéyaor sabhā-goer is defined as a "young man" (sabheyo yuveti), who is in his first age-span (prathamavayasī); "therefore [because he is a sabheya-], one who is in his first age-span is likely to become loved 200 by women". 201 The last statement is important in the light of what we know about the sexual activities in the sabhā, to which we will return later. The young age of the sabheya-, which is confirmed by some other texts, 202 is of interest here; this holds especially for the term prathama-or pūrva-vayasin-. As Harry Falk has pointed out, 203 the terms pūrva-, madhya-, and uttama-vayasá-(the first, middle, and last age-spans) are used elsewhere in ŚB (12.9.1.8; 12.2.3.4) to distinguish between three periods of a man's life: during the first of these, he still subsists on his father, though we may conclude from his frequenting the assembly hall and being attractive to women that he is considered an adult. Falk postulates that the term pūrvavayasá-pertains to the years immediately following the end of studentship or brahmacarya-; as this used to last between ages 8 and 16, the pūrvavayasá-would be the last years of adolescence, preceding the man's entering a settled life. This seems, as Falk points out (and here he is followed by Oberlies 204 ), to be the remnant of a very ancient system, closely paralleled in ancient Persia, 205 where the student years occurred between ages 7 and 15, followed by the years (15-20) as a "young man", yuvān mart (cf. yuvan-as a synonym of sabheya-), then adulthood and lastly old age. Furthermore, the years following brahmacarya-also coincide with the young man's living as a vratacārin-, vrātya-or sattrin-, which terms, as Falk's study in particular has made clear, denote the member of a sort of Männerbund, subsisting on cattle-rustling while studying the "esoteric" parts of Vedic lore in the wilderness, and frequenting the sabhā, where they indulged in (ritualistic) dicing, feasting, and promiscuous sexual relationships. This tallies more than well with Vasilkov's view of the sabhā-as a "men's house" for youngsters living in a Männerbund, though he doesn't seem to have taken notice of Falk's work on the topic.
Another important fact is that the sabheya-was always male; "men go to the sabhā, not women" (MS 4.7.4 206 ). The only women allowed in the assembly hall were courtesans. 207 According to Vādhūla Śrautasūtra 3.93, a woman who was "free to run off to the sabhā" was called sāhā-, and was considered impure from the "heat of mating" (mithunasya … śucā). That women who used to go to the sabhā were not considered respectable is clear also from other texts (e.g., ŚB 1.3.1.21); most well-known, albeit late, is Mbh 2.62.8-9, where, in the context of the fateful game of dice in the sabhā of the Kauravas, Draupadī is dragged into the assembly hall after having been staked and lost. There she complains, "What greater humiliation than that I, a woman of virtue and beauty, now must invade the sabhā? … From of old, we have heard, they do not bring law-minded (dharmyāḥ) women into their hall." 208 (Transl. van Buitenen, modified.) As protests are raised against her being forcedly led into the hall dressed in a single garment, Karṇa replies, "The Gods have laid down that a woman shall have one husband, scion of Kuru. She submits to many men and assuredly is a whore! Thus there is, I think, nothing strange about taking her into the hall, or to have her in one piece of clothing, or for that matter naked!" 209 (2.61.35-6; tr. van Buitenen.) Here it appears that a prostitute (bandhakī) -which Draupadī is considered to be, due to her having five husbandsis fit to bring into the sabhā, and that her slight clothing is only proper in this regard; what follows after these words of Karṇa is the famous episode where the Kauravas attempt to strip Draupadī naked, to make her humiliation complete.
The people frequenting the sabhā were thus, at least in Vedic times, young men, and courtesans. In this connection, mention should be made of ŚB 13.4.3.7-8, where the gandharvas and apsarases are represented, at the horse sacrifice, by "beautiful young men" and "beautiful young women" (yuvānaḥ/yuvatayaḥ śobhanā). We have already seen that yuvan-earlier in the same book of ŚB is given as a synonym of sabheya-. Intriguing is also the commentary on Śāṅkhāyana Śrautasūtra 16.2.11 (being a quotation from the ŚB passage), where the apsarases' being represented by "beautiful young women" is explained with the words, "Because no other (women; anyāsām being feminine) enter the sabhā" 210 (this being the scene of the sacrificial rite). The apsarases are thus, in the commentator's mind, defined by their being able to enter a sabhā; and this they have in common with the girls representing them at the ritual. The only reasonable explanation seems to be the view, prominent in the epic and classical literature, of the apsarases as celestial courtesans; a view which also seems to account, as Vasilkov suggests, for the presence of dancing and music-playing apsarases at the sabhās of gods like Indra or Brahmā, receiving the spirits of heroes slain in battle. 211 The evidence provided by Vasilkov for a connection between gandharvas and the sabhā is mainly based on materials from the epic, and thus late; but he also finds some support in Vedic texts. To begin with the epic part: Vasilkov notes a parallelism between the episode in the gambling hall in book 2, and the events at the court of Virāṭa in the beginning of book 4. In the latter episode, the exiled Pāṇḍava princes and their common wife live in disguise at the court of a foreign king; Draupadī, posing as a chambermaid, attempts to avert any sexual approaches by claiming to be the wife of five mighty, jealous gandharvas, but is nonetheless courted by the king's marshal, Kīcaka. When rejected, the enraged marshal chases her into the royal sabhā and beats her up (4.15.6ff). After this, Draupadī invites him to a nightly meeting in the royal "dancing hall" (nartanāgāra-), where girls learn dancing by day, but which is empty at night -yet there is a large bed there. When Kīcaka arrives in the dancing hall, he is met, not by Draupadī, but by Bhīma, the strongest of her husbands, who has been hiding in the bed; he kills Kīcaka by pushing his limbs into his trunk, a deed which Draupadī, when later questioned, attributes to her enraged gandharva husbands. Vasilkov suggests that the dancing hall, where erotic activity seems to take place at night, is to be compared to the "common dormitories" found in the young men's houses in many tribal societies; he also believes that the somewhat unusual term sairaṃdhrī-used for Draupadī's profession at the court, while meaning "chambermaid" in later Sanskrit, should be understood as a euphemism for the similar-sounding word sādhāraṇī-, "common (to several men)". (Note that the apsarases "are called 'common'" (sādhāraṇāḥ smṛtāḥ) according to Rāmāyaṇa 1.44.20, Baroda ed.) The wife of the five gandharvas would thus have been considered a prostitute, and her running away from Kīcaka into the sabhā, which she in book 2 considered it shameful for a woman to enter, would here be natural: "this time her appearance in the sabhā created no scandal, evidently the >>wife of the gandharvas<< might enter the sabhā without hindrance". 212 This should be compared to the episode in book 2, where her forced presence in the sabhā is defended with reference to her five husbands; she is thus considered a harlot.
Vasilkov finds an intriguing parallel to these two episodes in the Aitareya Brāhmaṇa, 1.27one version of the myth of the soma-barter. As the gods hand over the goddess Vāc to the gandharvas, she is designated as mahānagnā-, a "stark-naked one". This rather rare term found in Vedic texts is used for some sort of prostitute, who figures in the context of a few rituals. One of the so-called Kuntāpa hymns (AV 20.136; RVKh 5.22), of highly obscene content and believed to have been employed in some fertility rite, depicts the mahānagnī in the act of cohabitation with a male counterpart, the mahānagna-(mentioned only here). Presumably, the ritual context may have been comparable to the copulation act forming part of the mahāvrata rite, where a prostitute (puṃścalī-) had intercourse with a man from Magadha. 213 In the AiB account, the mahānagnā goddess is clearly given away to the, plural, gandharvas as an object of their common sexual enjoyment; she is thus comparable to Draupadī in the dicing episode of Mbh 2, who is, first of all, married to several men; second, is called a "whore" for this reason; and lastly, is dragged half-dressed into the sabhā, where the Kauravas then try to strip her entirely -make her "stark naked", like the mahānagnā. Draupadī's posing as the wife of several gandharvas in book 4 makes the parallelism to AiB 1.27 even more striking. And in both episodes, a sabhā -known to have been the scene of sexual activity between adolescent males and prostitutes -plays an important part. Vasilkov points out that there seems to be a connection between the mahānagnā and unmarried young women in the wedding-hymns AV 14.1-2; in the first of these, there is a prayer (vs. 35-36) for the bride to be bestowed with the splendor (varcas-) of "the hind-parts of the mahānagnā", as well as from liquor (surā-) and dice. The last two items are, as Vasilkov notes, prominent elements in the activity in the sabhā (as a gambling hall and scene of feasting); 214 and so are prostitutes. The blessing, he suggests, is meant to let the bride, who is leaving the adolescent life behind her, carry with her the best things of the sabhā, where she would previously have been among the girls serving the young men as concubines. 215 A connection between gandharvas and the assembly hall is also pointed out in the episode of Arjuna's visit to Indra's heaven in book 3, where the hero enters the god's sabhā, which is filled with singing, music-playing and dancing gandharvas and apsarases. At Indra's behest, Arjuna is further given dancing lessons by the gandharva Citrasena. Noting that dancing is considered a manly sport in many societies -no doubt even in the r̥ gvedic one, where we find 212 Ibid. p. 394. 213 Cf. Hauer 1927, pp. 274-78;Rolland 1972, pp. 64-7. 214 Vasilkov, p. 391; for surā-in the sabhā, cf. Falk 1986, pp. 89f. 215 Vasilkov, p. 393, writes that, "in the next hymn (AV. XIV.2) it is the apsarases who are asked to turn over to the bride their varcas", thus suggesting a parallelism between these beings and the mahānagnā. However, I have been unable to find such a prayer either in the hymn referred to or in the preceding one (14.1); at best, the gandharvas and apsarases are asked to be "good", not harmful, to the wedding procession, and Viśvāvasu is implored to return to his wives, the apsarases. warrior-gods like Indra and the Maruts depicted as dancers -Vasilkov suggests that the Mbh tale reflects an initiation into the "men's house", where the adolescent boys were taught the skills and knowledge befitting an adult man. When Arjuna later rejects the sexual invitations of the apsaras Urvaśī, and is cursed by her to be deprived of his virility, this is interpreted as a failed sexual initiation; in the "men's house", the boys would learn about sex through relationships with courtesans; by not taking part in these, they would not attain real manhood, and would be considered like impotent. According to Vasilkov, the promiscuous relationships associated with the sabhā would explain the concept of "gandharva marriages", which has been described above. He further suggests that the origin of classical Skt. gaṇikā-, "courtesan", is to be sought in the gaṇas ("hosts"), which is the common term by which the ancient sodalities or Männerbünde were referred to in Vedic and later literature; 216 a gaṇikā would, then, originally have been a woman "belonging to the gaṇa", as a common sexual partner shared by the unmarried young men.
While not suggesting that each and every trait in the conception of gandharvas and apsarases can be explained with reference to their connection with adolescence and the "men's house", I do believe such a connection to be well established, and a likely explanation for some of these traits. This holds for the frequent association between apsarases and dicing (this being one of the most prominent activities in the sabhā); some similar association on the part of the gandharvas might possibly be implied in the name of the "Kali gandharvas" (AV 10.10.13; JB 1.154-55). 217 As (decent) women were debarred from the gambling hall, it seems remarkable that the female apsarases are to be found there, presiding over the dicing. Considering the traditional conception of these deities as celestial courtesans, however, this fact becomes less remarkable. Promiscuous sexual activity was most certainly an established part of life in the sabhā; 218 and so were amusements such as dancing, feasting, and drinking. Telling is the association of the sabhā with the rare words naríṣṭ(h)ā-and narmá-in AV 7.12.2 and VS 30.6. The former passage -occurring in an invocation of the sabhā and the samiti-likely candidates when we look for inheritors of the máryas, are undoubtedly the vrātyas or "vow-takers". Long believed to have belonged outside the boundaries of Vedic culture proper, being associated with rites and customs considered impure by the authors of the sacrificial texts, the vrātyas are now 225 considered to have formed an important part of a Vedic agegroup system; according to Falk, they were Veda-students who had completed their brahmacarya-and now spent a few years in sodalities, living in the wilderness (araṇya-) and studying the "esoteric" parts (āraṇyaka-, the "wilderness-lore") of the sacred knowledge while subsisting on raiding neighbouring territories. (This lore included, i.a., the sattra and pravargya rites.) They were certainly closely associated with the sabhā (which was located in the wilderness, outside the village boundaries) and the ritualistic dice-game that took place there; 226 as well as to courtesans -those women who were allowed in the sabhā, first of all, but also -through the mahāvrata ritual -to the puṃścalī-or harlot (cf. AV 15.2) who performed an obscene dialogue with a brahmacārin, followed by ritualistic intercourse with a man from Magadha. Now, it is certainly interesting to find the gandharvas and apsarases, in the context of the human sacrifice, being assigned -a vrātya (VS 30.8,TB 3.4.5.5). Though such a direct connection is, to my knowledge, restricted to this passage, the (as I think, well established) association of the gandharvas with adolescence and the sabhā makes it difficult to reject it as simply ad hoc. That men's societies across the globe have been posing as representatives or embodiments of spirits or ancestors is well-known from ethnography, and this appears to have been the case also among several Indo-European peoples. 227 It may be noted in this regard that there is some kind of connection, pointed out by several scholars, 228 between gandharvas and the sage Keśin (or Baka) Dārbhya, likely to be identical with the "shaman" Keśin ("the long-haired one") of RV 10.136, who travels "along the course of the apsarases and gandharvas and the wild animals". This Keśin Dārbhya also has clear affinities to the vrātyas and the (with them closely associated 229 ) sattra sacrifice, 230 which he, for instance, teaches to the gandharvas and apsarases in MS 1.4.12, and which is in AVP 2.52.1 said to have been instituted by "the keśins". Indeed, the last verses of the vrātyastoma are called keśinīḥ according to JB 2.226, which is explained by the fact that vrātyas "keep themselves with long hair" (keśair iva hy ete caranti). 231 Elsewhere, Keśin has a group of followers called keśinīḥ (ŚB 11.8.4.6) or kaiśinīḥ (Vādhūla Śrautasūtra 4.37); and in older times these appear to have been called simply by the plural keśinaḥ, "keśins". 232 As Falk and others have noted, 233 long hair is a very common characteristic of young males in age-group societies, 234 and both brahmacārins and vrātyas let their hair and beards grow long. Long hair is also typical for the gandharvas, who, as we have seen, had their hair arranged into plaits. 235 In this connection, a passage from the Jain Sūyagaḍaṃga Sutta (1.7.10) is certainly interesting; as it happens, I came upon it only after finishing the above discussion. I quote here from Jacobi's translation: Men die as embryos, or as babies who do not yet talk, or who do so already; other men, as boys wearing five tufts of hair (paṃcasihā, Skt. pañcasikha-), or as youths, or in the middle age; at the expiration of their life all leave the body and die. 236 (Jacobi 2004(Jacobi [1895.) Pañcaśikha is, as we have seen, the name of a prominent gandharva (cf. already the śikhaṇḍín-of AV), and wearing five plaits is a characteristic also ascribed to apsarases. 237 In the passage quoted, pañcaśikha-clearly designates an adolescent boy; the commentator Śīlāṅka glosses it as kumāra-, "boy, youngling". A similar hairstyle seems to have been customary among the ancient Tamils, though in this case, amongst girls: "Many girls had their hair done into five plaits. After marriage the plaits gave place to a coiffure, known commonly as 'Koṇḍai'." 238

Concluding remarks
There are some further features of the gandharva mythology which are reminiscent of activities connected with the vrātyas -for example, the mahāvrata-or New Year festival, which is associated with them, 239 features music (e.g., from the lute or vīṇā), dance, ritualistic swinging, and cohabitation with a prostitute. 240 But a mere enumeration of common characteristics would be futile. What I think can be reasonably well established is a connection between gandharvas and the non-settled life of young males, characterized by activities such as dancing and feasting, and some which are frequently connected with the sabhā: most notably sexual licentiousness and dicing. As the celibate Veda-student would be unlikely to engage in such activities, the sabhā-attending youth, yuvān, who has completed his brahmacarya-but has not yet married and settled down, is the most likely candidate when we look for an earthly "counterpart" of the gandharvas (as the yuvānaḥ indeed are in ŚB 13.4.3.7). As the vrātyas clearly constituted at least a branch of this age-group, their association with gandharvas, though occasional, seems to be accounted for. As this was also the age when the esoteric parts of Vedic lore were studied in the wilderness, and sacrifices like the sattra-were performed, it would be tempting to connect this with the gandharvas' well-known knowledge of sacred mysteries (and especially rituals).
Summing up the results of this study, it has to be admitted that it is hard to find a single dominating trait that could explain the formation of the mythological being under discussion. The gandharva appears as a mediator (of knowledge, etc.) between heaven and earth; as guardian of the soma in heaven; as a mischievous spirit possessing women and causing insanity and miscarriages; as a fertility deity, etc. An attempt has, however, been made to show that many of these various traits and functions are interconnected; thus, the conception of the gandharva as beautiful, lusty and pleasure-seeking would be befitting a spirit of generation, while at the same time being closely connected with his possessing mortal women. At the basis of the Middle and Late Vedic mythology around gandharvas, it has been argued, is the exchange of sacral knowledge and substances for profane, especially sexual, enjoyment; this exchange takes on the one hand the form of possession of women, on the other, it is enacted in myths such as that of the soma-barter. A similar "intermediary" function of the gandharva, as noticed especially by Oberlies, could arguably explain a fair deal of the myths and rituals involving this kind of being: the transfer of soma to earth, of divine secrets to mortals, of pubescent girls to the married life. It may certainly be responsible for the gandharva's connection with rites of passage, such as weddings, and his being a sort of guardian deity of pre-married girls and, it seems, boys. Himself of youthful appearance and character, he appears to be especially connected with the "first" or adolescent life-stage (following childhood and studies, and thus a sort of "intermediary" stage before marriage and a settled life). This would also explain the gandharvas' and apsarases' association with the sabhā, with dice, with vrātyas and with prostitutes.
All this being said, it should be clearly stated that the subject is in need of deeper study; as is the entire, rather loose and confusing mythology around gandharvas and apsarases and its development through the ages. Hopefully, the present attempt will be a step in this direction. Aarne, Antti, andStith Thompson, 1981 [1951].