Memory, Desire, and “Magic”: Smará in the Atharvaveda

Abstract

This essay analyzes the interconnection between memory, desire, and verbal performance in the three so called “women’s love spells” in Atharvaveda 6.130–132. This study unpacks the many interconnected meanings of the term smará, which is used repeatedly in these poems, “memory”, “desire”, or “efficacious ritual speech”. I challenge the traditional definition of these texts as “magical” and argue that applying “magic” as an analytical category to ancient Hindu texts is deeply problematic. Instead, I propose that these poems are better understood in their historical and religious context as examples of ritual speech.

The Sanskrit noun smará has two distinct meanings, which at first glance appear quite unconnected to one another: “memory” and “desire”. The intriguing part, however, is that smará does not mean either memory or desire in Sanskrit, depending on the context in which the word is used; rather, it appears to carry both meanings simultaneously. To love, in Sanskrit literature, is to remember; without recollection there can be no desire. The complex intertwining between desire and memory is most strikingly illustrated in Kālidāsa’s classic play Abhijñānaśakuntalā (“The Recollection of Śakuntalā”, ca. 5th cent. CE). This intricate relationship between desire and recollection in classical Sanskrit literature has been examined in some detail by Charles Malamoud (Malamoud 1996) and Amanda Hunt (Hunt 2000). To date, there has been little systematic study of the oldest Sanskrit texts that fuse desire and memory, the so-called “women’s love-spells” of the Atharvaveda (AV 6.130–132). These three hymns are embedded in one of the four holy Vedas and have been memorized by generations of learned practitioners. I argue that these hymns shed considerable light on the conception of memory in the Vedic tradition as well as on the larger question of the relationship between speech and ritual power. This essay examines these Atharvaveda hymns in light of the texts’ own assertion of ritually powerful speech as something that engages with both memory and desire. In so doing, this study challenges the usefulness of the category of “magic” as applied to ancient Indian religious practice. I will demonstrate that classifying these hymns as “spells” and their contents as “magic”, as has often been done, is deeply problematic and that the texts are better understood within the framework of the Atharvaveda as powerful utterances that ritually transform memory and desire into social reality.

In the three hymns, Atharvaveda 6.130–132, which are only found in the Śaunakīya recension of the text, the narrator yearns for an absent beloved and hopes to evoke a similar erotic longing in him. These hymns are identified as part of a woman’s ritual in the Kauśikasūtras, a later appendix to the Atharvaveda, and by the commentators Keśava and Sāyaṇa. Many translators have therefore assumed that the hymns are love spells used by women to attract men (Bloomfield 1897, pp. 104–5; Whitney [1905] 1984, pp. 379–80; Griffith [1916] 1995, pp. 317–19). My focus here is not on establishing the gender of the speaker, for which the hymns themselves offer little in the way of direct evidence1, but rather on examining the ways in which remembrance and erotic love are intertwined in these stanzas.

Intriguingly, the term smará, which can variously be translated as either “desire” or “memory”, is here also used self-referentially about the hymns themselves: “O gods, send this smará forth. Let him yearn for me.” (AV 6.131.1). This usage of smará in the sense of a powerful verbal utterance is limited to Atharvaveda 6.131–133, and has not previously been examined by others investigating the complex memory-related terms in Sanskrit. In the following pages, I will analyze how this usage of smará in the sense of a ritual utterance, which at first sight seems highly idiosyncratic, is still intricately linked to notions of memory and desire. As we shall see, in these hymns, the verbal utterance itself is precisely a fusion of recollection and longing.

1. Smará and Its Cognates

Before analyzing these hymns of the Atharvaveda, however, we need to turn to the Sanskrit term smará itself. The Sanskrit noun is derived from the verbal root smṛ, “to remember”, and is related to other culturally significant Sanskrit terms like smṛti (“a recollected sacred text”), smaraṇa (“the act of recollecting”), and smārta (“a teacher well versed in the remembered oral tradition”). Numerous words in Sanskrit are formed from this root, their meanings ranging from the philosophical (smṛtipratyavamarśa, “the accuracy of a recollection”) to the erotic (smarastambha, “the pillar of love” or “memory stick”, i.e., “the male sexual organ”). Perhaps the act of passionate remembrance, in all its facets, plays such a significant role in ancient Indian culture precisely because of India’s rich history of orally transmitted texts and emphasis on memorization. To read a sacred Vedic text is, in one sense, meaningless; the text only assumes its true significance once it is committed to memory. To remember something is to relate to it in a deeply personal way.

The Sanskrit root smṛ has several cognates in other Indo-European languages. The reconstructed proto-Indo-European root *(s)mer- (“to remember”) is reflected in numerous Indo-European words related to recollection. The English words “memory”, and “remember” are both cognates, derived from Latin reduplicated forms memor, “mindful”, and memoria, “memory” (Pokorny 1959, p. 969; Partridge 1959, p. 396). A reduplicated form of *(s)mer- is seen in the name of the Norse giant Mimir, who guards the well of wisdom (Pokorny 1959, p. 969; Watkins 2000, p. 80). But we may note that the English word “mourn” (“to remember sorrowfully”) is also a derived from the same Indo-European root *(s)mer- (Partridge 1959, p. 397), as is the old Germanic mornen (“to grieve”) (Pokorny 1959, p. 969; Partridge 1959, p. 396) and the Old Norse morna, “to pine away” (Pokorny 1959, p. 969; Heggstad 1958, p. 471). The original meaning of “remembering” is here transformed into a sorrowful yearning for those lost, but still recalled. Other words derived from this root have similar shades of meaning: the Welsh marth means “sorrow” or “anxiety”, Cornish moreth “grief” or “regret”, Gaelic smur “sadness”, and Armenian mormok “regret” or “sorrow” (Partridge 1959, p. 396; Casey 2000, p. 353).

A similar sense of uneasy remembrance infuses the related Greek term μέριμνα (merimna “care”, “worry”). This word is attested in such Biblical passages as Matthew 13: 22: “And the one on whom seed was sown among the thorns, this is the man who hears the word, and the worry of the world and the deceitfulness of wealth choke the word, and it becomes unfruitful” and Luke 21:34: “Be on guard, so that your hearts will not be weighted down with dissipation and drunkenness and the worries of life, and that day will not come on you suddenly like a trap”. The memories of the world and the recollections of worldly life have here become so oppressive that they threaten to overcome the divine Word in the mind of the person who remembers.

The interrelatedness of memory, anxiety, and longing is not, then, unique to Sanskrit. While the meaning “erotic desire” is not attested in any of the manifestations of *(s)mer outside Sanskrit—with the possible exception, as we shall see, of the ancient Greek hímeros—the Sanskrit word smará reflects a sense of absence, which can also be seen in words like the English “mourn”. Mourning and loving may not be entirely distinct activities; they both involve a yearning for that which is remembered. Memory and a sense of loss conspire to form a longing that can manifest itself either as mourning or desire.

A possible parallel to the multifaceted Sanskrit term smára can be seen in the ancient Greek hímeros, “desire”. Although Weiss argues that hímeros is not, in fact, derived from the PIE root *smer (Weiss 1998, p. 50), I follow the majority of scholars in accepting the standard derivation of hímeros from PIE *smer, “to remember” (Frisk 1960, p. 726; Chantraine 1970, p. 464; Mayrhofer 1976, p. 549; Kloss 1994).2 Weiss distinguishes between hímeros and eros as desire of external and internal origin, respectively (Weiss 1998, p. 50). Hímeros arises in response to an external stimulus (a desire awakened by another), while eros originates within the desiring subject. Weiss points out that the “compulsive and external character of hímeros fits well with its use in contexts of love magic” (Weiss 1998, p. 50), an observation that is particularly relevant in comparison to the Sanskrit smará. The Greek hímeros is used in the context of Helen longing for her absent former husband, her former home, and her parents in Iliad 3.139–140. It also refers to Paris’ desire for Helen, spoken in her presence, but after she reviles him (3.446). Zeus’ desire for Hera (14.328 and 14.163) is associated with both remembrance and longing for an absent other.

We see in both smára and hímeros the notions of yearning and memory intertwined, which suggests that an undertone of longing for that which is absent may be present in the Indo-European root *(s)mer itself. To remember is a form of desire, and that desire has the potential to compel, as suggested by the use of both smára and hímeros in contexts that have been described as “love magic” (Kloss 1994; Faraone 1996, 1999). As I will demonstrate in this paper, however, “magic” is an inadequate hermeneutic category to describe these texts’ intertwining of memory, desire, and the power of ritual speech.

2. Atharvaveda 6.130-132: An Analysis

One of the most haunting expressions of the weaving together of memory and desire in Sanskrit literature is found in the hymn Atharvaveda 6.130:

(1)

rathajítāṃ rāthajiteyínām apsarásām ayáṃ smaráh/

dévāḥprá hiṇuta smarám asaú mám ánu śocatu//

(2)

asaú me smaratād íti priyó me smaratād íti/

dévāḥ prá hiṇuta smarám asaú mám ánu śocatu //

(3)

yáthā máma smárād asaú námúṣyāháṃ kadá caná/

dévāḥ prá hiṇuta smarám asaú mám ánu śocatu //

(4)

ún mādayata maruta úd antarikṣa mādaya/

ágna ún mādayā tvám asaú mám ánu śocatu //

(1)

This is the memory–desire (smará) of the Apsarases,3 the ones who conquer chariots, belonging to those who conquer chariots.4 O gods, send this memory–desire (smará) forth. Let that man burn for me.5

(2)

Let him remember (smaratāt)6 me, let my beloved remember (smaratāt) me. O gods, send this memory–desire (smará) forth. Let that man burn for me.

(3)

That he may remember (smarāt) me, but I him never again, O gods, send this memory–desire (smará) forth. Let that man burn for me.

(4)

Madden him, winds, and madden him, air! Madden him, fire! Let that man burn for me.

In this hymn, we see erotic desire (smará) depicted as an intricate interplay between remembering (smará) and forgetting. The narrator, rhetorically constructed as female, longs for a remembered beloved, who appears to have forgotten her and lost his desire for her. Her desire is anchored in memory, but her feelings are unrequited and her one-sided remembrance has therefore become a form of suffering. Not only does she call for her beloved to remember (smṛ) her; her utterance also turns into a prayer for merciful oblivion: Let her forget, while he still remembers and yearns. Intriguingly, the hymn refers to itself as a smará as well; the speaker’s memory and desire have turned into a powerful verbal utterance, the hymn itself. The term smará operates at several levels of meaning at once here, which it why it is so challenging to translate. The term evokes both erotic longing and remembrance, but also the very language in which these emotions are captured. There is no word in English that captures all of these meanings, which is why I have used the awkward “memory–desire” here, perhaps more precisely rendered as “memory–desire–utterance”.

Both Bloomfield and Whitney translate smará simply as “love” in these stanzas of the Atharvaveda (Bloomfield 1897, pp. 104–5; Whitney [1905] 1984, p. 379). While the term certainly carries that meaning as well, these translations ignore the vital connection the term has to memory and utterance as well as desire. Recognizing that smará is a meta-level reference here to the very hymn in which the word is embedded, Ludwig renders smará as “Liebeszauber” or “love-charm” (Ludwig 1878, p. 515). Griffith similarly translates smará in this hymn as “love-spell” or “spell” (Griffith [1916] 1995, pp. 317–19). These translations recognize that smará is not just a term for desire, but a term applied to the very hymn itself, although the translation “spell” is still problematic, as I will demonstrate later in this essay.

In these stanzas, smará is not just something (a memory) that connects the mind and heart to the external world, but something that reaches out and takes control over something outside of itself. The utterance, like the memory, provides access to that which is temporally, spatially, or emotionally distant. The hymn becomes an articulation of both desire and absence, but most importantly, a vehicle for overcoming that absence.

It is worth noting that the text’s narrator is calling upon several supernatural entities to ensure the efficacy of the utterance. While the Apsarases, who are semi-divine female beings, are associated with erotic love in general, some of the other deities invoked in this hymn are not: “Madden him, winds, and madden him, air! Madden him, fire!” (AV 6.130.4). The speaker is not merely calling upon the elements of nature here but invoking powerful Vedic gods such as the Maruts (winds) and Agni, the ritual and cosmic fire, and the gods (dévāḥ) in general. The invocation of the gods, so common in Vedic literature, places the hymn firmly within the Vedic liturgical context and the domain of the Vedic priest.

In his reading of a Ṛgvedic hymn that includes a “female” speaker, the “Weapons Hymn” (Yuddhasūkta) of Ṛgveda 6.75, Whitaker argues that the union of man and woman (the warrior and his feminine weapons) represent the symbolic coupling between a male poet–priest and the female speech (Whitaker 2019b, p. 139). He argues that the priestly poet, through the hymn’s imagery, “underscores his authoritative training, his right to speak liturgically, and the powerful nature of his feminized voice” (Whitaker 2019b, p. 143). Similarly, in his analysis of ā́yus (“life”) in the Ṛgveda, Whitaker draws on Bourdieu to show that ā́yus functions as a discursive formation that “works to construct and circumscribe identities and practices of ritual participants of the early Vedic ritual tradition” (Whitaker 2011, p. 42). Whitaker critiques Malinowki’s approach to ritual as something that has “magical” efficacy and proposes that the ritual is not about wish fulfillment, but rather a strategy to discursively strengthen the ritual specialists’ social capital and reaffirm the power of the ritual itself (Whitaker 2011, p. 47). I argue that smará does similar work in AV 6.130–132 in asserting the power of the Atharvaveda poet’s liturgical speech, a power that even extends over men’s and women’s desires and memories.

When we turn to the next hymn, Atharvaveda 6.131, we see that the term smará again becomes a reference to a form of speech. Like the preceding poem, this hymn is infused with a sense of longing for that which is absent. Again, we see that the hymn itself, like the memory of the absent lover, purports to be able to overcome both emotional and physical distance. The multifaceted smará as memory–desire–utterance is not merely a recollection of the past, but a merging of then and now, of the absence and the presence, a dissolution of all distance:

(1)

ní śīrṣató ní pattatá ādhyò3 ní tirāmi te/

dévāḥ prá hiṇuta smarám asaú mám ánu śocatu //

(2)

ánumate ’nv idáṃ manyasvákute sám idáṃ námaḥ/

dévāḥ prá hiṇuta smarám asaú mám ánu śocatu //

(3)

yád dhávasi triyojanáṃ pañcayojanám áśvinam/

tátas tváṃ púnar áyasi putráṇāṃ no asaḥ pitá //

(1)

Pangs of longing I draw down upon you, from head to foot. O gods, send this memory–desire (smará) forth. Let that man burn (śocatu) for me.

(2)

Assent to it, O Assent! Guide it, O Intention! O gods, send this memory–desire (smará) forth. Let that man burn for me.

(3)

If you run three leagues away, or five leagues, or as far as a horse can run, you shall come back again and be the father of our sons.

The term smará is again used self-referentially in the sense of an utterance in the following hymn, Atharvaveda 6.132, a text that reaffirms the liturgical context of that utterance:

(1)

yáṃ deváḥ smarám ásiñcann apsv àntáḥ śóśucānaṃ sahádhyá/

táṃ te tapāmi váruṇasya dhármaṇā //

(2)

yáṃ víśve deváḥ smarám ásiñcann apsv àntáḥ śóśucānaṃ sahádhyá/

táṃ te tapāmi váruṇasya dhármaṇā //

(3)

yám indrāṇί smarám ásiñcad apsv àntáḥ śóśucānaṃ sahádhyá/

táṃ te tapāmi váruṇasya dhármaṇā //

(4)

yám indrāgnί smarám ásiñcatām apsv àntáḥ śóśucānaṃ sahádhyá/

táṃ te tapāmi váruṇasya dhármaṇā //

(5)

yám mitráváruṇau smarám ásiñcatām apsv àntáḥ śóśucānaṃ sahádhyá/

táṃ te tapāmi váruṇasya dhármaṇā //

(1)

The memory–desire (smará), glowing (śóśucānaṃ) with longing (ādhί), which the gods have poured into the Waters, that I heat for you by Varuṇa’s7 ordinance.

(2)

The memory–desire (smará), glowing with longing, which all the gods have poured into the Waters, that I heat for you by Varuṇa’s ordinance.

(3)

The memory–desire (smará), glowing with longing, which Indrāṇī8 has poured into the Waters, that I heat for you by Varuṇa’s ordinance.

(4)

The memory–desire (smará), glowing with longing, which Indra and Agni9 have poured into the Waters, that I heat for you by Varuṇa’s ordinance.

(5)

The memory–desire (smará), glowing with longing, which Mitra10 and Varuṇa have poured into the Waters, that I heat for you by Varuṇa’s ordinance.

Here, the speaker’s unrequited desire is interpreted as a form of heat, and the utterance capturing that yearning is ritually equated with the fire offered to the gods in the Vedic sacrificial rite. The verb śuc (“to burn, to glow, to blaze, to suffer”) is used in both hymn 6.131 and 6.132 to capture a sense of both desire and suffering, but also to place those emotions within a Vedic ritual context. The verb śuc and its derivatives are often used to characterize Agni in the Ṛgveda. In ṚV 6.15.7, Agni is called śúcim pāvakám puró adhvaré dhruvám, “the blazing pure one in front, steadfast at the ceremony” (Jamison and Brereton 2014, p. 789). Agni is likewise associated with blazing in 8. 44.21: agníḥ śúcivratatamaḥ śúcir vípraḥ śúciḥ kavíḥ śúcī rocata áhutaḥ, “Agni, best possessor of flaming commandments, flaming inspired poet, flaming sage poet, flaming he shines when he is bepoured” (Jamison and Brereton 2014, p. 1118). Rhetorically, then, the use of the verb śuc to characterize the desire about to be kindled in the absent lover in AV 6.131, and the speaker’s own smará in AV 1.132, transforms the personal emotions into ritual entities under the control of the poet–priest.

The underlying conceptual metaphor here is one that is common cross-culturally, that of desire being a form of fire, or in the language of cognitive linguistics: desire is fire (Kövecses 2010, p. 4; Jurewicz 2019, p. 42). But this conventional metaphor has a deeper meaning in the context of Vedic ritual texts: desire is not only fire in general, but sacrificial fire, which implies that this desire can be mapped to certain kinds of liturgical power. The metaphorical mapping between the source domain (fire) and the target domain (desire) is a complex one in these hymns of the Atharvaveda; fire is both a physical element and a significant part of the ritual. The conceptual metaphor desire is fire co-activates two different but interrelated source domains: fire as a source of physical heat, and fire as central element in the Vedic sacrifice. This “conceptual blending” of two metaphorical domains (Fauconnier and Turner 2002) gives the conceptual metaphor more depth in the Vedic context than the same metaphor might have in other cultural settings.

In these verses, smará encapsulates multiple meanings of the Indo-European root *(s)mer- all at once: smará is the speaker’s anguished memory of her beloved, her desire for him, and the charm that fuses her recollection and her yearning into a powerful utterance intended to awaken a similar memory and longing in the other person. In this case, we cannot say that memory is a metaphor for desire or vice versa; these two meanings of smará are only separate in translation, but fused into a single meaning, “memory–desire”, in the Sanskrit language. It is possible, however, to read the other usage of smará as a powerful verbal utterance, which only attested in AV 6.130–132, as a conceptual metaphor, memory–desire is an utterance. In this case, the two domains of memory–desire and utterance are mapped to each other conceptually in multiple ways: An utterance, like a memory or desire, can bring close that which is absent in space or time, and a memory or desire can—at least in these three hymns of the Atharvaveda—be a powerful ritual element, like a liturgical utterance.

Although it is possible to interpret this hymn as an author’s personal expression of yearning, we should not overlook the strong ritual elements present in the hymn. The utterance is “poured into the waters” and then “heated”. This description evokes the Vedic hymns to Apáṃ Nápāt (“The Child of the Waters”, see Ṛgveda 2.35), who glows as a burning fire within the heavenly waters; in this hymn, the utterance itself becomes a metaphorical “Child of the Waters”, a deity associated with creation (2.35.2) and verbal eloquence (2.35.1). The repeated mention of Varuṇa, the Vedic guardian of the cosmic order, suggests that the AV hymn, far from a mere personal utterance, has both ritual and cosmic implications. The liturgical legitimacy of the poet–priest’s actions is affirmed by the repetition of the phrase “by Varuṇa’s ordinance.”

At first sight, this usage of smará in the sense of a spoken utterance is unique to these three hymns of the Atharvaveda. It is tempting to speculate, however, that this notion of a powerful verbal utterance as the locus of memory and desire may also be haunting the more well-known concluding stanza of the Īśā Upaniṣad:

• vāyur anilam amṛtam athedaṃ
• bhasmāntaṃ śarīram.
• Oṃ krato smara kṛtaṃ smara
• krato smara kṛtaṃ smara.
• The never-resting is the wind, the immortal.
• Ashes are this body’s lot.
• Oṃ, mind, remember the deed, remember!
• Mind, remember the deed, remember!
(Īśā Upaniṣad 18)

The injunction in this stanza to remember one’s past deeds at the end of one’s mortal life is usually interpreted simply as an invitation to look back and reflect on what one has done with one’s life. But the use of smara (“remember”, here a verb in the imperative form) may also suggest that through the very utterance of this verse itself, a past deed is brought near,11 in the same way a reluctant lover is brought back through a spoken verse in the Atharvaveda. Remembering, and speaking of that which is remembered, blurs the past and the present, and creates a powerful cosmic shortcut to that which is recalled.

3. “Magic” in the Atharvaveda

The three brief hymns of memory (smará) of the Atharvaveda translated here contain rich linguistic, literary, and ritual layers. The diction is emotional, abrupt, and fragmented, but these texts also invoke the main guardians of the cosmic order among the Vedic gods and evoke complex cultural and religious notions of the power of memory and the efficacy of the spoken word. Is it reasonable to classify these utterances as a form of “magic”, a term many Western scholars have applied to the sometimes bewildering array of poems and hymns about disease, healing, wealth, cows, crops, love, hate, safety, demons, and danger contained in the Atharvaveda? 19th century scholars such as Bloomfield saw in the Atharvaveda an appalling contrast to the serene and pious prayers to the gods in the Rigveda; to Bloomfield, the Atharvaveda is “not at all squeamish in the choice of its themes, and exhibits the ordinary Hindu not only in his aspect of devout and virtuous adherent of the Brahmanic gods, and a performer of pious practices, but also as the natural semi-civilized man; rapacious, demon- and fear-ridden, hateful, lustful, addicted to sorcery” (Bloomfield 1899, p. 61). Bloomfield here expresses not only an orientalist prejudice against the culture he was studying, but also a view of magic as an earlier evolutionary stage than religion, a view grounded in the work of Herbert Spencer and James Frazer.12 Modern scholars will distance themselves from Bloomfield’s pejorative characterization of the Atharvaveda and of Vedic practitioners, but the distinction between religion and magic still lingers in recent work on South Asian religions.

Although many non-Indian scholars have applied the term “magic” to South Asian texts and practices (Goudriaan 1978; Gonda 1978, p. 104ff; Glucklich 1997), recent scholarship has challenged this projection of a Western analytical category onto ancient Indian texts (Stutley 1980, p. xii; Whitaker 2011, 2019a). I argue that a Western distinction between religion and magic, which is problematic enough in itself,13 becomes quite meaningless when superimposed on ancient Sanskrit texts, where no such distinction exists. Whereas many Western scholars have regarded the Atharvaveda with suspicion and dismissed its contents as “sorcery” or “magic” as opposed to the pure “religion” of the Ṛgveda, the Indian tradition itself does not express a similar anxiety over the contents of the fourth Veda, nor does the tradition distinguish between the religion of the Atharvaveda and that of the other Vedas. As Stutley observes in her study of ancient Indian “magic”: “Essentially, magic and cult are the same, since all rites are basically magical, any difference being that of the method used to influence or propitiate the unknown powers” (Stutley 1980, p. xii). Jamison and Witzel make a similar observation: “In general, it is difficult and misleading to separate “magic” from other sorts of ritual activity” (Jamison and Witzel 1992, p. 49).

The hymns in Atharvaveda 6.130–132 do have a ritual context according to the Kauśikā Sūtras 36.13-14, which describes an accompanying ritual consisting of making an effigy of a desired man, throwing beans at its head, and throwing lighted arrows all around it (Stutley 1980, p. 53). But there is nothing in the Vedic tradition itself to suggest that the rituals described in the Kauśikā Sūtras fall into a different category from other Vedic ceremonies often classified as “religious”, such as the agniṣṭomá. Any differentiation between “magic” and “religion” is completely absent in the Vedic tradition itself.

In his seminal work Making Magic: Religion, Magic, and Science in the Modern World, Styers argues that “magic” as a modern analytical category has grown out of a particular (Western, colonial, Protestant) cultural desire to define rationality and modernity, and that “magic” itself has functioned for many scholars as a “foil for modernity” (Styers 2004, p. 8), a constructed “other” against which both religion and modernity can be defined. Tambiah likewise argues that the distinction between “religion” and “magic” was a Protestant legacy taken over by Victorian theorists like Tylor and Frazer, who gave “magic” a universal significance as an analytical category in order to further a particular Western European worldview (Tambiah 1990).

But does the troubled historical legacy of the term “magic” render it invalid as an analytical category? Both Braarvig (1999) and Dubuisson (2016) have made valiant attempts at rehabilitating “magic” as a scholarly category, while acknowledging the term’s problematic colonial and theological history. Braarvig differentiates between three forms of usage of the term “magic”: intratextual, intertextual, and extratextual. At an intratextual level, something is defined as “magic” by the practitioners themselves. Such intratextual references to “magic” can be found in ancient Greek texts, as well as among modern Neopagans. An intertextual usage of the term “magic” involves someone accusing others of practicing “magic” and using it as a term of condemnation, a usage that is found both in ancient Greece and in colonial-era works describing the practices of those who are culturally “other”. An extratextual usage of the term “magic”, according to Braarvig, involves a scholar using “magic” as an analytical category, in a “disinterested” way, even if the practitioners whose rituals are analyzed do not use that term themselves. Braarvig defends the use of “magic” as an analytical category by arguing that a distinction between “magic” and “religion” is not merely a modern construct, but a more universal dichotomy that can be traced back to the works of several ancient authors cross-culturally, such as Augustine (Braarvig 1999, p. 41) and the 6th century CE Mahāyāna Buddhist philosopher Bhavya (Braarvig 1999, p. 45). Braarvig defines magic, following Frazer, as ritual actions that are coercive and performed for personal, rather than collective purposes (Braarvig 1999, p. 52; cp. Frazer 1911–1915, pp. 222–25) and proposes that “magic” can still be useful as an analytical category provided that it is used without judgment.

Dubuisson, likewise, attempts to restore the academic study of magic (or rather, of multiple “magisms”) from “the shadow of religion” (Dubuisson 2016, p. 181). Dubuisson traces the reification and separation of “religion” and “magic” to early Catholic writers and examines the upholding of those categorical boundaries as a (Christian) theological project (Dubuisson 2016, p. 108). Nevertheless, he proposes that “magic” still has value as an analytical category if divested of its negative associations and emancipated from the religion–magic binary. If “magic” is no longer coded as a contrast to “religion”, what remains of this concept? For Dubuisson, “magisms” or “magical processes” are ritualized words and actions, often focused on the personal, practical, and mundane, performed by human agents in the context of a “highly symbolic” world (Dubuisson 2016, p. 140), ruled by “sympathetic associations” (Dubuisson 2016, p. 142). Dubuisson’s definition of “magisms” echoes Malinowski’s classical definition of magic as practically oriented rituals performed to allay anxieties over things that remain outside human control (Malinowski 1948, p. 14). Both Braarvig and Dubuisson contrast rituals that serve an immediate and practical purpose for the individual with those that do not and propose the former must be classified as “magic”.

But is this distinction applicable to Vedic texts? Even if it were possible to divest the category of “magic” of its pejorative connotations and its colonial history, does “magic” as an analytical category give us further insight into the Vedic material? Here, I think the answer is negative. Not only is there no evidence of a conceptual distinction that corresponds to “religion”/”magic” in the Vedic texts themselves, but viewing the Vedic material through the hermeneutic lens of “magic” obscures some crucial parts of the internal coherence of the liturgical world of the Vedas. By viewing the smará hymns of the Atharvaveda as categorically different from the Ṛgveda’s poems to Indra or Varuṇa, we risk losing sight of the fact that these texts do very similar things and are embedded in similar ritual contexts.

4. Smará as Verbal Performance

Although these Atharvaveda hymns are composed in the first person, it is not reasonable to assume, as many scholars have done, that they represent one individual’s personal experience. The ritual context suggests that the hymns are not a mere outpouring of personal emotion following a bad break-up, but rather a religiously significant text embedded in a particular ritual context. The narrative of the hymn itself is framed by a larger text that relocates the utterance from the realm of the personal to the ritual. Here, we can observe a contrast with the so-called ancient Greek “love magic” inscriptions, whose context suggests a personal, rather a ritual use. While the wording of the Atharvaveda hymns is not radically different from the 1st century lead tablet inscription from Carthage that reads: “ … take away the sleep of that woman until she comes to me and pleases my soul… lead [blank space] loving, burning on account of her love and desire for me…”,14 the context is very different indeed: the lead tablet is meant to be used by one person in secret to win another’s affections, while the Atharvaveda hymns are transmitted by generations of priests as part of a sacred corpus of texts and used in priest-led rituals.

It is unlikely that the use of a feminized speaker’s voice in the Atharvaveda hymns implies that the poems are composed by a woman (see footnote 1). Rather, the use of a feminine speaker in these hymns could be interpreted as a discursive strategy that underscores the power of the (male) ritual specialist. In this reading, the hymns are less about a woman’s power over a man than about subordinating a man’s free will and desires to the ritual power of the priest speaking on behalf of a female client. Whitaker has demonstrated that a patriarchal Vedic ideology is rooted in ideals of masculine strength embodied either in the physical strength or warriors or in the ritual power of male priests (Whitaker 2016). When these Atharvaveda hymns appear to show a man subjected to the desires of a female speaker, this does not undermine that patriarchal ideology; rather, it strengthens the idea that the ritual power is the dominant one. This significant point is lost if the hymns are read as “magical” in the sense of personal, practical, and coercive.

But if the Atharvaveda hymns are not personal, why do they employ the first person? First person speech is very common in Vedic texts and used extensively from the Ṛgveda onwards. This use of the first person must be understood as performative rather than personal. As Caley Smith has shown, it is particularly fruitful to examine Vedic oral texts in the context of performance theory (Smith 2019). Smith theorizes a “performance grammar” of Vedic texts, comprising three axes of textual reference: spatial, temporal, and personal; texts that repeatedly refer to “here”, “now”, and “I” or “we” are often meant to be part of a ritual performance. In the case of these Atharvaveda hymns, we can identify several such performance markers in the text: “this is the utterance…”, “this utterance…”, and the first person pronoun “I”, all of which should signal to the reader that the text needs to be understood in the context of an oral performance. In fact, several Vedic texts insist that their words are meant to be performed, spoken out loud:

• ná nau mantra ánuditāsa eté
• máyas karan páratare can´āhan
• These thoughts, if unuttered,
• will not bring us joy even on a distant day
(Ṛgveda 10.95.1, translation from Jamison and Brereton 2014, p. 1549).

In his insightful study of orally performed texts, Lars Lönnroth analyzes how a text’s form and content are contingent on the situation in which it is performed (Lönnroth 1978, p. 7). Lönnroth refers to the place where the text is performed as the stage (“scene” in Swedish), and argues that an oral performance involves a dual stage (“dobbla scenen”), a sort of narrative double exposure where the performed text exists both as an internally coherent narrative in itself and as a performed narrative that interacts with the audience (Lönnroth 1978, p. 9). I argue that the smará hymns of the Atharvaveda occupy such a dual stage in that their contents must be understood simultaneously as an internally cohesive narrative (in this case, about a female speaker who wants to evoke desire in a man who appears disinterested) and as a ritual text meant to be performed. Here, I find Smith’s proposed distinction between three levels of ancient Indian texts, based on a schema outlined in Śāṅkhāyanagṛhyasūtra 1.1.2.3–5, particularly useful: the adhidaiva level pertaining to the divine, the adhyātma level that pertains to the personal, and the adhiyajña level, which pertains to ritual performance (Smith 2019, p. 3; cp. Jamison 2019, p. 64). In the smará hymns of the Atharvaveda, it is easy to identify the text’s adhidaiva level as relating to the gods, an adhyātma level as a narrative of personal suffering, and an adhiyajña level as text intended for ritual performance. This threefold reading of a text as theological, personal, and ritual/performative may indeed be more helpful overall than a distinction between texts that are “religious” or “magical”. To classify the Atharvaveda hymns as “magical” is to dismiss both the adhidaiva and the adhiyajña levels of the text in favor of a narrow reading of the hymns as adhyātma, or relating to the personal. While the hymns may fit certain traditional European definitions of “magic”, in that they may be read as an attempt to force a particular outcome for personal ends, this reading ignores the theological and ritual aspects of the texts. It is perfectly possible to read these Atharvaveda hymns as a depiction of one individual’s desire to make someone love her back, but this reading does not take into account that these are also poems about the Vedic gods and the cosmic order, and that the narrator’s desire is articulated in the context of that larger cosmic order. A reading of the hymns as individual and “magical” also does not take into account the texts’ performance markers and known performative context. The hymns’ inclusion in a textual corpus faithfully transmitted by generations of trained male priests also speaks against an interpretation of the hymns as a form of individual magic. While the hymns articulate a narrator’s desire, those desires are not at all at odds with the larger social values of the time; the culmination of the narrator’s utterance is a plea for the reluctant lover to return and become the father of sons, a desire that has both personal and social implications.

I propose that the recognition of the three textual layers—the theological, the personal, and the performative/ritual—is useful in the study of religious texts far beyond the study of Vedic religion. An attention to these hymns adhiyajña level helps us re-frame the long-standing question of “magic” in ancient India. Many acts that have been classified as “magical” by previous scholars, such as verbal utterances or ritual actions that are meant to effect a change in the external world, are performative in nature (Tambiah 1973, p. 199). Ritual utterances, such as those in Atharvaveda 6.131–132, are perlocutionary acts; they are not mere assertions but rather a form of speech that purports to do something, an important distinction Austin notes in his seminal work How To Do Things with Words (Austin 1962, p. 6; see also Tambiah 1973, p. 220). The very uttering of the words is itself a “leading incidence in the performance of the act” (Austin 1962, p. 8). A ritual utterance is similar to the English phrase “I now pronounce you man and wife” in that the speech act itself “does something”, instead of simply narrating something that happened. But just like the statement “I now pronounce you man and wife” is only regarded as valid in a particular context (spoken by a person with the social and religious authority to perform marriages, spoken to partners who have agreed to marry), so a Vedic ritual utterance must be spoken in the correct context, by an authorized person, and at the correct time, in order to be socially valid. As Austin writes of such speech acts: “There must exist an accepted conventional procedure having a certain conventional effect, the procedure to include the uttering of certain words by certain persons in certain circumstances” (Austin 1962, p. 26). In other words, a speech act is only regarded as effective if there is a social consensus that makes it so by virtue of its context. A perlocutionary ritual statement is not only an articulation of a desired change in the world, but also an assertion of a particular social power on the part of the speaker, a power granted by the speaker’s society that renders the uttered words capable of effecting that change.

But perlocutionary speech is not limited to ancient Indian texts that have been classified by Western scholars as “spells” or “magical formulas”; the vast majority of Vedic literature, including both the hymns of the Ṛgveda and the so-called “spells” of the Atharvaveda, falls into the larger category of perlocutionary speech, speech that asserts the power to change some part of reality. I argue therefore that a distinction between “magic” and “religion”, a distinction arising from a European Enlightenment-era privileging of forms of practice that closely resembled Protestant Christianity over all others, is far less useful when applied to ancient Indian text than Smith’s distinction between a text’s adhidaiva, adhyātma, and adhiyajña levels.

Attention to the adhiyajña level of the smará hymns of the Atharvaveda may further help us understand the complex interrelationship between memory and desire in these texts. I argue that memory is itself theorized in these hymns as a form of performance. When the speaker pleads with the gods to “send this smará forth” (6.130.1–3; 6.131.1–2) in order to cause desire in the beloved, smará is no longer mere personal recollection, but a ritual entity with the power to effect change. The pronoun this (ayam) indicates that this smará is the speaker’s utterance itself. The idea of an utterance “going forth” and being effective in the world is attested in numerous Vedic texts (Ṛgveda 3. 51.2; 5.87.1; 8.43.1, etc.). Why is this utterance specifically referred to as a smará (memory–desire) here? The association between remembrance and utterance is of course essential in Vedic culture, where texts are transmitted orally. Without memory, there can be no text and no utterance. All verbal performances rely on memory, but these hymns add to this the insight that memory is itself a form of desire for that which is absent, a bringing near of what is temporally, spatially, and emotionally distant. At the text’s adhidaiva level, what is brought near are the gods, at the adhyātma level a reluctant lover, and at the adhiyajña level the result of a successful ritual performance. In a larger sense, then, the smará hymns of the Atharvaveda are a comment on the very nature of text in ancient India: ritual performance fuses memory and desire, presence and absence, into a powerful verbal utterance. Rather than classify some Vedic texts as “magical” and others as “religious”, it makes more sense to read the hymns of both the Ṛgveda and the Atharvaveda as examples of ritually effective speech.