In this essay, I wish to address the problems a historian encounters while explaining the function and origin of ancient rituals. One is particularly confronted by these problems when dealing with a ritual such as the Navarātra. With regard to its function, the festival resists sharp distinctions between the sacred and the temporal because it simultaneously propitiates a deity and solemnizes the authority of a ruler. It seems to be two things at the same time: a rite of religious power and a rite of political power. In fact, in the Southern Navarātra, for instance as celebrated in Vijayanagara, the worship of the Goddess would take place largely out of view in a private shrine, while all the individual rites of the festival appeared to publicly celebrate a cult of the king in the larger communal area: the Navarātra thus appeared, as it did to Portuguese and Persian visitors to the Vijayanagara court, to be a political festival with a minor religious dimension.¹ While explaining its origins, we run into even greater difficulty as the earliest traces of the Navarātra are found in more than one distinctive religious tradition: the Vaiṣṇava, the Śaiva, the Purāṇic and even possibly in regional traditions of communities outside mainstream ‘Hindu’ traditions. Where it truly “originates” is therefore difficult to see, though for clarity’s sake I have proposed here that the Vaiṣṇava domain was where a mature, theologically coherent² conception of the rite evolved. On the whole the ritual appears to have been of a composite character at each stage of its manifestation. The overall impression is that we are looking at many permutations of different rites with different origins that attached themselves around the central figure of the Goddess, and through her, and the demon-slaying mythologies surrounding her, acquired a structural and thematic unity. In the following, I shall present and probe these ambiguities of function and origin through a scheme of the richly varied regional traditions of the Navarātra that emerged in the course of its history.

Navarātra, the autumnal festival of the warrior goddess Caṇḍikā, is today one of the most eagerly anticipated events of the Hindu calendar. Built up over nine lunar days and culminating on a tenth, the festival fulfills several, apparently disparate, purposes: it offers obligatory worship to the Goddess, without which her wrath could become implacable (so legends warn in dire tones); wards away omens from—and thereby symbolically cleanses and renews—a community of people; and bestows the ritual stamp of victory on the military forces of a kingdom. What is particularly noteworthy even in the modern ceremony is the symbolic connection between the political, the martial, and the religious, manifested by a priest through a sequence of meaningfully choreographed rites and staged within a lavishly ornamented arena of worship publicly open to all. Facilitating this connection is the Goddess herself, who intertwines in her being an image of secular rulership and transcendent, or spiritual, sovereignty. Little, though, is understood about the historical reasons for the culmination of this overlap in the figure of Durgā and its ritualized realization in the ceremony of the Navarātra. Doubtless, the Navarātra, given that it was dedicated to a war-goddess, played a significant role in preparations undertaken by a medieval kingdom to wage war, and furthermore, to affirm social structure, as notable studies in the past by Alexis Sanderson (2007, 195–311), Shingo Einoo (1999, 33–70) and Ralph Nicholas (2013) have shown. The season of autumn, which in many cultural traditions, and also in classical India, was when armed campaigns would take place, must also have formed a reason for the presence of military rituals such as the lustration of weapons and war-animals during Durgā’s Navarātra: following the monsoon, during which it is notoriously difficult to make journeys, the autumn, when the skies are clear and the weather cooler, formed the perfect season to venture forth on campaign.³ Doubtless there must also have been an association between social governance, its urban political locus, and the Goddess, as there is in the modern ceremony. But when did the autumnal festival acquire such a role in sanctifying heroic endeavor, and in affirming roles and functions dispersed within the social organization? For in many Navarātras certain groups and lineages were traditionally associated with particular rituals: for example the priest’s duty, the cutting of the head of the animal, the making of the effigy, or the provision of virgin girls for worship, etc. This still remains at the forefront of any basic inquiry into the nature of this ritual.

However, answering this question presents certain methodological complications. Literature (primarily in Sanskrit) indicates that during the period the festival developed and was popularized throughout South Asia, viz. the 5th to 12th centuries CE, it grew into a locally diverse tradition. At present these regional traditions seem on the surface to be but tenuously interrelated, and in their diversity forestall our entertaining the possibility of there having been common templates of origination.

Apart from the Navarātra’s multiplicity of form, other factors have prevented, it would seem, a full history of the festival from being undertaken—apart from, it is important to note, Einoo’s (1999) pioneering study, “The Autumn Goddess Festival Described in the Purāṇas.” These factors are as follows: difficulty in interpreting and evaluating sources; confusion prompted by the presence of non-Brahmanical rituals within an outwardly Purāṇic-Brahmanic ritual framework; and ambivalence in status because of the important roles played by people outside the caste-system in the ritual sequence. However, these difficulties, confusions and ambivalences are not insuperable, and do in fact point to an important characteristic of the ritual: that its position within either the Brahmanical or the non-Brahmanical realms was never very clear. Both sides claimed certain aspects of the ritual as theirs, and in fact operated in tandem within its domain. A political synergy between different power-groups was effected through the course of the festival, as indeed has been recently shown by Nicholas (2013) and which will grow even more evident through surveying the different traditions.

However, if we look at a wide range of ritual descriptions in Sanskrit contained in Purāṇas for which the conjectured dates seem reliable, in their reused forms in Dharmaśāstric compendia (nibandhas) and in ritual manuals (paddhatis), together with ethnographic accounts of ceremonies, where available, a historical pattern begins to emerge. The full analysis of that pattern is treated in my recent book, Heroic Shāktism: the Cult of Durgā in Ancient Indian Kingship, including texts and translations from the relevant ritual descriptions.⁴ Since it may also be useful to offer a succinct overview of the historical pattern, here I offer a condensed summary of that larger narrative, presented in schematic form below. I also take the opportunity here to include three very early sources that I did not have the opportunity to consider for the arguments made in my book: the Southern Cilapattikāram,⁵ and two passages from the Mahābhārata and the Kādambarī. These offer fresh insights int0—if not the Navarātra proper—the ritual background of the Goddess: her worshippers, their provenance and the purpose for which she may have been worshipped prior to the development of the liturgical materials on the Navarātra. The consideration of the Southern material in particular leads to re-conceiving the historical development of the Goddess’s worship south of the Vindhya mountains. In the book, I had suggested that

The Deccan seems to have followed in the wake of the eastern form of the Navarātra outlined in the Devī[purāṇa] and the Kālikā[purāṇa] until at least the early half of the 14th century [on the basis of passages from those works appearing in Deccan Dharmaśāstric nibandha-literature] … The gradual independence of the southern tradition and its advocacy by the 15th century of a Navarātra that was qualitatively different from the eastern tradition in that it celebrated Daśamī differently and eschewed rites that were Tantric in their tone are attested by the eyewitness accounts of the Navarātras of the Vijayanagara kingdom, of Mysore under the Wodeyars, of Ramnad and Śivagaṅgai in Tamil Nadu.

Sarkar 2017, 258–259

On the other hand, the Cīlapattikāram suggests that already in the early centuries of the common era, a local form of worshipping the Goddess for power in battle, like the Navarātra, was celebrated in the Tamil country, and that this included possession, trance and bacchanalia. On this basis, it is possible to suggest, first, that long before descriptions of Durgā’s worship appeared in materials of an Eastern provenance, she was popular in the south and, second, that the direction of liturgical influence could have been the other way: the ritual of bacchanalian enjoyment offered to a goddess of battle could have entered into Eastern liturgies, in which they occupy a prominent place, from Southern prototypes. Moreover, all three textual examples are magnificently composed, and exemplify all that is most vivid and energetic in poetry about the Goddess.

In some of our earliest sources, the rite did not begin in autumnal Āśvina, the month usually associated with the Navarātra. Rather it began in the monsoon month of Śrāvaṇa, and prior to that seems to have been a popular festival celebrated by everyone regardless of sectarian affiliation. It then came to replace a more established set of Brahmanical military traditions (such as the worship of weapons and the lustration of the army) practised in Āśvina. Once these Vedic royal traditions were harmonized with the worship of the Goddess, they altered in character to become a goddess-centred heroic tradition, in which sanguinary rites to calm down Durgā’s fiery nature began to dominate. But further alterations in the character of the ritual followed, as it was incorporated by other religious “specialisms.” One of the most critical transformations to have occurred in the structure of the Navarātra is the appearance of Tantric rituals in descriptions of the rite emerging from East India, notably from Orissa and the kingdom of Mithilā. Compared to the military festival of the earlier Vaiṣṇava and then the Vedicized formats, these rituals amplify the power-bestowing efficacity of the ritual by including rituals that grant siddhis (powers). Moreover, there is no single goddess, but many, and of many forms, names and natures. In literature from Mithilā, the rite expands to nine days to include, apart from the worship of nine forms of Durgā of different colours, an array of Tantricized rituals such as the purification of elements (bhūtaśuddhi), the worship of the sixty-four yoginīs, the installation of mantras in the body (nyāsa), self-identification with the deity (a ritual that, although also found in pre-tantric materials, came to be associated with tantric practice⁶), rites bestowing powers (siddhi) held at midnight, and the heightening of the Goddess’s personality so that her ferocious properties are thought to take over. She is invoked as Kālī-Lauhadaṇḍā (Kālī, goddess of the iron rod) in mantras in the mediaeval Bengali rite. In the Maithila rite she is even summoned as Cāmuṇḍā into a bel branch, which is then worshipped as the vehicle of her essence throughout the duration of the worship. In Orissa, as shown by Sanderson (2007), mantra elements from the Kashmirian Kālīkula were incorporated into the Mahānavamī traditions of Bhadrakālī. The effect of this Tantricization was the enhancement of the power-bestowing agency of the ritual, desirable no doubt for rulers eager to achieve victory in the battles they were about to undertake.

The sources in which the above ritual patterns are described are as follows (the specific references with emended Sanskrit texts and analyses are to be found in the locus indicated in the accompanying footnotes):

I. Early Vaiṣṇava phase in the monsoon⁷

Harivaṃśa 57.35–36; Mahābhārata 4.5.29 ff. and 6.22.6 ff., old Skandapurāṇa 60.46; Kādambarī pp. 30–31; Harṣacarita p. 126; Caṇḍīśataka 16; Gaüḍavaho 318, 319, Purāṇic citations in Dharmaśāstric compendia from Mithilā and Bengal.

II. Incorporation with a Brahmanical military festival in Āśvina¹⁰

Viṣṇudharmottarapurāṇa 2.158.6cd–7, Agnipurāṇa 267.13cd–16ab (repeating Viṣṇudharmottara); Varāhapurāṇa cited in the Kṛtyaratnākara, pp. 364–365.

III. Expansion and inclusion of Tantric power-rituals in Eastern Court Traditions¹¹

Devīpurāṇa, Kālikāpurāṇa, Kṛtyakalpataru, Durgābhaktitaraṅgiṇī, Durgāpūjātattva, Durgāpūjāviveka, Bhadrakālīmantravidhiprakaraṇa in Sanderson (2007); account of the Durgā Pūjā in Kelomal, West Bengal (Nicholas 2013).

IV. The Southern and Western Court Traditions¹²

Caturvargacintāmaṇi, Sāmrājyalakṣmīpīṭhikā, Puruṣārthacintāmaṇi, accounts of ceremonies in Śivagaṅgai and Ramnad, Tamil Nadu (Price 1996), Portuguese traveller accounts from the Vijayanagara Empire (Stein 1983).

To the above sources, I would like to add a passage from the Mahābhārata (first noticed and pointed out to me by Sahiṣṇu Bhaṭṭācārya, Bardhaman, West Bengal in a personal communication), whose importance in regard to the worship of Nidrā, Durgā’s early form, requires emphasis. This passage appears in the Sauptikaparvan (Mahābhārata X, 8.64–68) and suggests that the Vaiṣṇava Nidrā, goddess of Sleep and Death, presided over and blessed battle as a dangerous spirit (kṛtyā). Called Kālarātri, identified with apocalyptic destruction, adorned with a peacock feather (śikhaṇḍinīm) that evokes her alliance with her brother Kṛṣṇa, as prevalent in this period,¹³ she manifests herself when Aśvatthāman, the son of Droṇa, secretly enters the Pāṇdava camp and goes on a murderous rampage. Hers is a strange, menacing apparition:

kālīṃ raktāsyanayanāṃ raktamālyānulepanām |

raktāmbaradharām ekāṃ pāśahastāṃ śikhaṇḍinīm ||

dadṛśuḥ kālarātriṃ te smayamānām avasthitām |

narāśvakuñjarān pāśair baddhvā ghoraiḥ pratasthuṣīm ||

harantīṃ vividhān pretān pāśabaddhān vimūrdhajān ||

svapne suptān nayantīṃ tāṃ rātriṣv anyāsu māriṣa |

dadṛśur yodhamukhyās te ghnantaṃ drauṇiṃ ca nityadā ||

yataḥ pravṛttaḥ saṃgrāmaḥ kurupāṇḍavasenayoḥ |

tataḥ prabhṛti tāṃ kṛtyām apaśyan drauṇim eva ca ||

tāṃs tu daivahatān pūrvaṃ paścād drauṇir nyapātayat |

trāsayan sarvabhūtāni vinadan bhairavān ravān ||

Mahābhārata X, 8.64–68

Good sir, they saw her, Kālarātri, standing, smiling, alone, blue-black in hue, with red mouth and eyes, garlands and unguents of crimson, red robes, a noose in one hand, a peacock feather [in her hair], binding men, horses and elephants with her horrifying fetters while she stood, capturing many headless ghosts trapped in her noose, leading those asleep in their dreams to other Nights (rātriṣv¹⁴ anyāsu). And at all times the best soldiers saw the son of Droṇa slaughtering. From the time when the battle between the Kuru and Pāṇḍava armies began, they saw [both] that evil spirit (tām kṛtyām) and the son of Droṇa. The son of Droṇa later felled those who had first been struck by this divinity [Kālarātri], terrorizing all creatures while shouting out ferocious bellows.

Towering over the nightmarish battlefield, a grinning image of Death, Kālarātri governs both sleep and death, ensuring the interceptor certain triumph during his secret raid.

The Tantricization of Durgā’s worship must have been well established by 700 CE, by which time it must have already been correlated with worship of Śiva, rather than Viṣṇu. Such is the impression created by a minutely detailed description of Durgā’s shrine and her worshipper in Bāṇa’s Kādambarī (pp. 224–228¹⁵). On his way from Hemakūṭa to Ujjayinī, the hero of the work, Candrāpīḍa, stops for shelter at a shrine of the Goddess that comes midway. The shrine is nestled in the midst of a densely wooded forest, and the narrative, through a telescoping of perspective from outside to inside the shrine, provides a leisurely description of its design, appearance and atmosphere. From afar Candrāpīḍa first sees a “crimson ensign,” “inscribing the sky with a gold trident, from which swung a terrifying bell making a raucous clanging (gharghararava) that dangled down from an iron chain attached to the tip, arranged with a yak-tail whisk as splendid as a lion’s mane” (dolāyitaśṛṅgasaṅgilohaśṛṅkhalāvalambamānagharghararavaghoraghaṇṭayā ca ghaṭitakesarisaṭāruciracāmarayā kāñcanatriśūlikayā likhitanabhaḥsthalam … raktadhvajam; p. 224). Going ahead a little, he then sees that the Goddess Caṇḍikā “was enclosed by a door made from the ivory of wild elephants, as yellowish-white as fragments of ketakī filaments, and an iron architrave (toraṇa) bearing an ornamental garland of black iron mirrors surrounded by a row of red yak tail whisks resembling a garland of Śabara heads horrific with tawny hair” (ketakīsūcikhaṇḍapāṇḍureṇa vanadviradadantakapāṭena parivṛtāṃ lohatoraṇena ca raktacāmaraparikarāṃ kālāyasadarpaṇamaṇḍamālāṃ śabaramukhamālām iva kapilakeśabhīṣaṇāṃ bibhrāṇena … caṇḍikām; pp. 224–228). Then he notices the dvārapāla (guardian of the gate), about which it is said that “[Caṇḍikā] had protected her entrance with an iron buffalo installed in front, which, because of the fact that it had been marked by palms [dyed with] red-sandalwood, seemed to have been stamped by Yama’s hand-prints red with blood, the red eyes of which were being licked by jackals greedy for drops of blood” (sanāthīkṛtadvāradeśām abhimukhapratiṣṭhena ca vinihitaraktacandanahastakatayā rudhirāruṇayamakaratalāsphāliteneva śoṇitalavalobhalolaśivālihyamānalohitalocanena lohamahiṣeṇa; Kādambarī, p. 224). Then through the main entrance, the temple yard: “Her courtyard was adorned with thickets of red aśoka trees, the spaces between the branches of which were made gapless by flocks of perching red cockerels, [trees] which appeared to reveal unseasonal clusters of blooms in their fear” (śākhāntarālanirantaranilīnaraktakukkuṭakūlaiś ca bhayād akāladarśitakusumastabakair iva raktāśokaviṭapair vibhūṣitāṅgaṇāṃ; p. 225). (More in fact is said about the overflowing mass of flowers, trees and even lion cubs that populate her front courtyard, slippery with blood.) Then the portal to the sanctum sanctorum, a riot of colour and form: “She was being illuminated by the entrance, on which there were hanging cloths reddened by lamp-smoke, a row of bracelets made of peacock-throats festooned [over it], a garland of bells closely-set and pale with powdered flour-cakes, which supported two door-panels, [studded] with tin lion heads with thick, iron pins in their centres, barricaded with an ivory-rod bolt, carrying [what seemed to be] a necklace of sparkling bubbles that were mirrors oozing yellow, blue and red [light]” (avalambamānadīpadhūmaraktāṃśukena grathitaśikhigalavalayāvalinā piṣṭapiṇḍapāṇḍuritaghanaghaṇṭāmālabhāriṇā trāpuṣasiṃhamukhamadhyasthitathūlalohakaṇṭakaṃ dattadantadaṇḍārgalaṃ galatpītanīlalohitadarpaṇasphuritabudbudamālaṃ kapāṭapaṭṭadvayaṃ dadhānena garbhagṛhadvāradeśena dīpyamānāṃ; ibid.).

Then follows the image of the Goddess, which in its association with the terrible, and in the predominance of the colour red, matches the conception of Kālarātri in the passage from the Mahābhārata: “She was installed on an altar of black stone” (adhyāsitāñjanaśilāvedikām, Kādambarī, p. 224). “Her feet were never bereft of cloths [dyed with] red lac thrown upon the mound of her seat [on the altar] as if they were the lives of all creatures arrived there for shelter; she resembled an inhabitant of the Underworld because of the intense darkness obstructed [only] by the flashes from axes, spears, etc., weapons deadly for beings, that seemed to hold nets of hair stuck from decapitations because of the reflections of black yak-tail whisks cast [upon their surfaces]; she was adorned in garlands of bilva leaves furnished with gleaming fruits and buds anointed with red sandalwood, that were like hanging garlands of infant-heads; she expressed cruelty with limbs worshipped with clusters of kadamba flowers ruddy with blood, which horripilated, it seemed, at the thrill of the flavour of the keen roar of drums during the animal-offering; she bore the coquettish apparel of a woman going out to meet Mahākāla at night, with a vine-like body furnished with a raiment reddened with saffron-dye, with a face with red eyes, whose brows were furrowed into a frown, whose lip was crimsoned with betel that was blood, whose cheeks were reddened by the light shed from ear-ornaments of pomegranate flowers, with a forehead on which there was a tilaka dot of vermillion made by a Śabara beauty, covered by a magnificent gold turban (cāmīkarapaṭṭa). She was worshipped by goats … mice … antelope and black serpents … She was praised on all sides by flocks of old crows.” (piṇḍikāpīṭhapātibhiś ca sarvapaśujīvitair iva śaraṇam upāgatair alaktakapaṭair avirahitacaraṇamūlāṃ patitakṛṣṇacāmarapratibimbānāṃ ca śiraśchedalagnakeśajālakānām iva paraśupaṭṭiśaprabhṛtīnāṃ jīvaviśasanaśastrāṇāṃ prabhābhir baddhabahalāndhakāratayā pātālanivāsinī, ivopalakṣyamāṇāṃ raktacandanakhacitasphuratphalapallavakalitaiś ca bilvapattradāmabhir bālakamuṇḍaprālambair iva kṛtamaṇḍanāṃ śoṇitatāmrakadambastabakakṛtārcanaiś ca paśūpahārapaṭahapaṭuraṭitarasollasitaromāñcair ivāṅgaiḥ krūratām udvahantīṃ cārucāmīkarapaṭṭaprāvṛtena ca lalāṭena śabarasundarīracitasindūratilakabindunā dāḍimakusumakarṇapūraprabhāsekalohitāyamānakapolabhittinā rudhiratāmbūlāruṇitādharapuṭena bhṛkuṭikuṭilabhruṇā raktanayanena mukhena kusumbhapāṭalitadukūlakalitayā ca dehalatayā mahākālābhisārikāveṣavibhramaṃ bibhratīṃ … chāgair … ākhubhir … kuraṅgair … kṛṣṇasarpair … ārādhyamānāṃ sarvataḥ kaṭhoravāyasagaṇena … stūyamānām; pp. 225–226).

A Draviḍa ascetic, portrayed as a comical figure, is said to be her priest; perhaps Bāṇa was conscious in making the priest a Draviḍa, on account of the widespread worship of the goddess Koṛṛavai, later correlated with Durgā, in the South? There are apparently several Tantric rites that Bāṇa pejoratively associates with the priest: he, “the ageing Draviḍa religious man” “demeans Durgā with his prayers for the boon of sovereignty over the Southern lands” (dakṣiṇāpatharājyavaraprārthanākadarthitadurgeṇa … jaraddraviḍadhārmikeṇa; p. 226); “he had copied a hymn to Durgā on a strip of cloth” (paṭṭikālikhitadurgāstotreṇa; ibid), “he had collected palm-leaf manuscripts of spells, Tantras and jugglery the letters of which were written in red lac and fumigated with smoke” (dhūmaraktālakākṣaratālapattrakuhakatantramantrapustikāsaṃgrahiṇā; ibid); “he had written down the [work known as ] the ‘Doctrine of Mahākāla’ instructed to him by a withered Mahāpāśupata mendicant” (jīrṇamahāpāśupatopadeśalikhitamahākālamatena, pp. 226–227); “he was one in whom the disease of talking about [finding] treasure had arisen” (āvirbhūtanidhivādavyādhinā, p. 227); “in him the wind [disease] of alchemy had grown” (saṃjātadhātuvādavāyunā, ibid.); “he entertained the deluded desire of becoming the lover of a Yakṣa maiden” (pravṛttayakṣakanyakākāmitvamanorathavyāmohena, ibid.); “his collection of practices for mastering mantras for invisibility had grown” (vardhitāntardhānamantrasādhanasaṃgraheṇa, ibid.); “he was acquainted with a hundred tales about the marvels of the Śrīparvata mountain” (śrīparvatāścaryavārttāsahasrābhijñena, ibid.); “his ear-cavities were punched by those possessed by piśāca-demons, who had run to him when struck by white mustard seed he had empowered with mantras more than once” (asakṛdabhimantritasiddhārthakaprahitapradhāvitaiḥ piśācagṛhītakaiḥ karatalatāḍanacipiṭīkṛtaśravaṇapuṭena ibid.); and “he had used magic powders for snaring women many times on aging mendicant ladies, who having arrived from other lands retired [there to rest]” (anyadeśāgatoṣitāsu jaratpravrajitāsu bahukṛtvaḥ saṃprayuktastrīvaśīkaraṇacūrṇena).

While it would be imprudent to treat this example of poetic literature as a bald record of fact, it is possible to see Bāṇa’s extensive and richly crafted episode of this horrific, yet magnificent, temple as a reflection of social attitudes to the Goddess and her worship. There is a mix of suspicion, fear and reverential awe underlying the image of the forbidding shrine tucked away in the wilds, with its Tāntrika priest who knows not how ‘appropriate’ worship should be conducted, and its blood-spattered, grisly interiors. The very opposite of this ambivalent attitude surfaces in Bāṇa’s unequivocally laudatory poem to Durgā, the Caṇḍīśataka—verse 8 of which is consciously alluded here in “she seemed to be scolding the wild buffalo who had offended by moving the trident-shaft by scratching his shoulders [on it]” (skandapīṭhakaṇḍūyanacalitatriśūladaṇḍakṛtāparādhaṃ vanamahiṣam iva tarjayantīm; Kādambarī p. 226). The topos of Mahiṣa scratching his back on the post appears in Caṇḍīśataka 8 too,¹⁶ in which there is a mischievous pun with the word sthāṇu that means both “post” and “Śiva.” The saviour of Dharma in the Caṇḍīśataka contrasts with the menacing though beauteous figure here. One may suggest that the wider context for this attitude of ambivalence is a historical transition: the Goddess first imagined as we have seen as a kṛtyā in the Mahābhārata was being absorbed within mainstream devotional practices, through which her demonic attributes became ‘toned down,’ balanced by the benevolent and the charming, but nevertheless remained, at the stage of Bāṇa’s compositions, tinged with a degree of the terrible. The comic portrayal of the priest registers the fact that by Bāṇa’s time Durgā’s worship had acquired firm cultural associations with Śaiva tantric rites. There was also an association of her site of worship with wild environments inhospitable to people, to flora and fauna in general (note that most of the ornaments in her shrine are of plants and flowers) and even, it seems, with a peculiarly Southern religious attitude.

Regarding the Southern Navarātra, it is tempting to conjecture that the roots of worshiping the Goddess in Devagiri and Vijayanagara drew also upon the older cult of Koṛṛavai, the stag-borne goddess described in the old Tamil poem, the Cīlapattikāram.¹⁷ In fact many salient elements of Durgā’s rituals in general (especially, though, in the East), such as the transactional nature of worship, trance, possession, ecstatic dancing, singing hymns, the important role of women, virgin-worship and heroic self-sacrifice involving blood are to be found even here, which suggest that among all elements of the Navarātra, these appear to be the earliest. In Canto XII Koṛṛavai is said to be worshipped by cattle-raiders for victory in their missions.¹⁸ The canto, called “Vēṭṭuvavari” (The Hunter’s Song), portrays the Eyinar community worshipping their protective goddess for victory before setting off on a raid. The chapter describes vividly the stages of pūjā at the shrine of Aiyai (Koṛṛavai), eulogized throughout the canto as Durgā, the slayer of Mahiṣa, the sister of Viṣṇu and the consort of Śiva. First, a respected Maravar lady Śālinī, an oracle, becomes possessed and dances, singing a hymn urging the hunters to offer tribute to the Goddess.¹⁹ In the hymn Śālinī rebukes the men for growing weak and no longer robbing passers-by,²⁰ the implication being that the Goddess will re-invigorate them with heroic zeal. After the oracle performs, a virgin is selected from the Eyinars, in what appears to be an early form of the kumārīpūjā, and treated with especial care as the Goddess. Dressed in tiger skin, with a snake of silver and a wild hog’s tooth in her matted hair, a necklace of tiger-tooth, a bow of wood and seated on a stag, the kumārī is brought before the shrine of the Goddess, set in lush and verdant groves of fragrant and flowering trees.²¹ The women offer her various gifts of dolls, beautiful birds, paints, scents, food and flowers with much fanfare and the beating of drums.²² After worshipping the Goddess, the virgin goes into a trance, and speaks to the heroine of the poem Kaṇṇakī, introducing her to Koṛṛavai, who from this moment in the ritual, it is suggested, becomes a living presence.²³

Koṛṛavai appears. She bore a moon on her hair, a third eye on her forehead; her lips were red, her throat blue with poison like that of her consort Śiva. The snake Vāsukī was her girdle and she wore a bodice resembling snake-teeth, an elephant’s hide over her upper body and a tiger skin over her hips; she carried a trident. There are rich ornaments on her feet. Dark in hue as a sapphire,²⁴ bejewelled, youthful, beautiful, ascendant on the head of the buffalo demon, she is called, among many names, the sister of Kṛṣṇa, Durgā, Gaurī, the giver of victory, worshipped by Viṣṇu and Brahmā,²⁵ and also the defeater of Kaṃsa.²⁶ The names and descriptions indicate that even at this early period, circa 450 CE, when the poem is thought to have been composed,²⁷ the Goddess, whose initial sectarian affiliation was with Kṛṣṇa, had already become associated with Śiva, and moreover had acquired an independent identity as a supreme divinity, worshipped by all the gods. Another hymn is sung by a girl to the virgin dressed as the Goddess, in which the duality of Koṛṛavai-Durgā is emphasised in a series of rhetorical questions or paradoxical contrasts: she is worshipped by gods and is an exalted repository of Vedic knowledge, yet also stands on a buffalo head adorned with wild animal hides; standing as light above the trinity of Viṣṇu, Śiva and Brahmā, she stands also on a humble stag with twisted black horns, holding aloft, with hands adorned with delicate bangles, a cruel sword; she who is consort of Śiva with three eyes also has a fierce red-eyed lion and the Vaiṣṇava conch and the discus.²⁸ She is also said to have danced the marakkāl, a dance on wooden legs, to defeat demons. If anyone invokes her wearing a victory garland (veṭci) before setting forth to seize cattle, omens of defeat will appear in the enemy’s village, and the Goddess will accompany the hero on his quest before his bow.²⁹ In the song, the plenitude of captured cattle is then praised, and the Goddess is asked to accept the raiders’ blood offered by cutting their necks to her in thanks.³⁰ This offering of flesh and blood is described as the Goddess’s price for the victory conferred on the warrior and outlines the transactional nature of the worship.³¹ The hymn becomes hypnotic at this climactic moment of blood-offering as in verse after verse the Goddess is asked to accept the blood.

Four things illuminated by the description are worth pointing out. Koṛṛavai is already treated as an eclectic deity merged with Durgā, herself a cluster of Vaiṣṇava and Śaiva elements. From the context here it is tempting to speculate that Durgā as a deity of royal power and kingship may trace her roots from local warrior communities such as the Eyinār, who sustained themselves by periodic looting necessitating armed confrontation. In fact, the evidence of the old Tamil text prompts us to refine the schematic diagram, and suggest that the first phase of the ritual going back to the early centuries of the common era was widely practised in the South and, along with blood sacrifice, included ecstatic singing, dancing and deity-possession. When this archaic phase of worship is described in literature, Durgā has already ceased to be a peripheral deity, as I had assumed she still was at this time when I wrote my book, and is considered both the consort of Śiva and Kṛṣṇa’s sister, and empress of the gods.

A historical survey of the Navarātra’s developmental pattern reveals that it acquired its characteristic shape from a confluence of two different traditions: first, a festival of the goddess Nidrā-Kālarātri on the ninth tithi of the dark half of a monsoon month, centred on sanguinary rites exhibiting heroism, and second, a military tradition celebrated in autumn centred on lustrating weapons and armies with fire to ward away ill omens. In this way, through a gradual process of coalescence and subsequent transformation, the Navarātra acquired two of its hallmarks: the central place of the Goddess as the deity to whom all the rituals were dedicated, and the autumnal season as its most favourable and appropriate time. The phases by which this confluence occurred and then further developed show us how a relatively small single-day civic festival, performed by Vaiṣṇavas as part of the celebrations of Janmāṣṭamī, gradually expanded into a much longer rite thought to safeguard society and the political class, which then came to be performed in the month of Āśvina. This development paralleled the attenuation of the rite’s importance for the Vaiṣṇavas and its absorption, first by Śaivas, who would promote the worship of the Goddess on Navamī by fasting, in literature about rites for lay devotees,³² and then by the more widespread Purāṇic-Brahmanical tradition. These sectarian absorptions provided impetus to the popularity of the rite among rulers of upcoming kingdoms eager to cultivate the ritual apparatus of goddess-sanctified kingship.

We can map out two broad regional traditions to have matured later in courts: an Eastern and a Deccan one. By and large these formed the basic blueprint for localized variations. As descriptions of Rajput rites of the śamī tree and the weapon-shooting on Daśamī from colonial ethnographic reports from the nineteenth century show, many of the Rajasthani royal rituals were extremely similar to the template of the Southern Navarātra.³³ The most splendid Navarātras seem to have flowered in the ornately ritualized Tantricized environments of Eastern India, among which the kingdom of Mithilā provides us with the most detailed testimonia. These appear to have percolated (in as much as this is reflected through citations) into traditions as far afield as the Deccan. In Mithilā the ten-day structure seems to have matured, and the ferocious identity of the Goddess took a central place in the Tantric rites of Mahāṣṭamī and Mahānavamī. This North-Eastern tradition developed into the Eastern or Gauḍīya tradition—a trend evidenced by a fifteenth-century Bengali work, the Durgāpūjātattva by Raghunandana, which incorporates rituals of a Bengali character, such as summoning goddesses and worshipping them in nine leaves from crops. A separate, even more markedly Tantric tradition, with elements borrowed from the Kubjikā cult, developed later in the Navarātra or Dasain of Nepal, but for the time being this will remain excluded from the discussion. In the South, rituals of a Tantric character were largely eschewed. The Navarātra was choreographed around the public display of the king, his court, his weapons and war-animals and magnificent parades, while the worship of the Goddess, and indeed the summoning of the Goddess into the king, occurred privately. During these days the king would worship nine virgins considered vessels of nine forms of the Goddess (different from the Navadurgās of Mithilā and Nepal) for powers such as mastery over enemies, knowledge, riches and an abundance of slaves and slave-girls, as described in ritual instructions in Sanskrit. The most resplendent examples of this version of the rite took place, it seems, in the Vijayanagara empire.

In this way what we find are many rituals of a large-scale, communal, public character clustering and growing according to differing political environments (as the Navarātra was chiefly promoted by the court) around the figure of Durgā. This process allowed a more sophisticated ritual interaction to develop between political agents (chiefly the ruler, then the army and the polis) and the Goddess (and her forms), who in the course of the Navarātra’s transformations cements her role as the deity who grants the goals of kingship (military victory; territorial protection) and protects communal areas such as fortresses, citadels and palaces. However, in spite of its proliferation one element remained the cornerstone of the rite: blood sacrifice. A key aspect of the nature of the ritual as a pact between the Goddess and solicitor of rewards, this remained constant throughout the development of the Navarātra and indeed even today is seen to be critical to its success. It is possible that before its appearance in Vaiṣṇava sources, the worship of Nidrā-Kālarātri in the monsoon was a widespread popular festival of heroism based on blood-sacrifice, including ecstatic communal bacchanalia, that could have formed part of a marauder’s cult, as in the worship of the stag-riding goddess Koṛṛavai. It was gradually absorbed into the influential sectarian traditions when goddess-cults came to be elevated during the Gupta period, as inscriptions from Valkhā, Madhya Pradesh, in the late Gupta period attest.³⁴

Recently, the historian Kunal Chakrabarti (2001) has suggested that the origins of the Navarātra lie in indigenous practice, and that its late emergence in Sanskritic literature is the culmination of a long process whereby the Goddess was gradually brought into the Sanskritic sphere. The Goddess, Chakrabarti argues, is a strategic means whereby peripheral and popular deities and traditions can be absorbed into the mainstream. Her festival was the time of the year when these popular traditions could be made public and shown to cohere around her.

Indeed this was the case, and such is also made evident in the description in the Cīlapattikāram. The festival of the Goddess, unlike its Vedic autumnal ancestor, integrated rites of different affiliations. Apart from the Tantric, other rites, performed by indigenous groups, would regularly be incorporated into the ritual sequence. In the Cīlapattikāram we are shown that, while the main community profiting from the worship are the Eyinār hunters, a representative of another group, the Maravars, plays a critical ritual role as the oracle. Certain social groups, for example, were authorized to carry through the animal sacrifice. In this sense the Navarātra united disparities within the social canvas in which it was embedded. The obligatory performance of rites that would involve everyone regardless of their caste represents a social inversion that the Navarātra set into motion during the classical period. It was at this time that the strict hierarchies enforced by the orthodox social order were overturned, albeit for a limited period, as the single day Śābarotsava attests. This leads us to question the long-held assumption that the Navarātra is a Purāṇic festival. Indeed, although in outward character it was, since it was taught in texts that were Purāṇic, and since it was further elaborated on by Sanskrit writers working to strengthen the Brahmanic order, it nevertheless was elusive in essence. I would argue that this lack of affiliation was one of the chief reasons why it grew into the most important ritual of political and communal affirmation. That it was one of the few rituals of elevated, that is to say Sanskritic status, that solidified the status and place of outcaste groups, and publicly displayed subversive rites that would otherwise have been deemed suspicious by brāhmaṇas, such as the caste-dissolving, orgiastic śābarotsava (The Festival of Śabara-tribes) on Daśamī taught in the Kālikāpurāṇa and Dharmaśāstric literature, served to identify it as a ritualized act of cohesion.³⁵ The Goddess herself was a metaphor of this cohesion, worshipped by both outcastes and people within the caste hierarchy. Indeed, literature, particularly classical kāvya, shows that her role as an outcaste deity preceded that of the Goddess of special importance to a kṣatriya. The importance for kṣatriyas is emphasized in the Devīmāhātmya, from perhaps the late-eighth century CE. Though images of the Goddess in the presence of warriors offering their blood to her appear from as early as the seventh century, it is in this work that we are first presented with what became a canonized narrative of the Goddess blessing a kṣatriya king and a vaiśya merchant, thereby being firmly associated with the power-model of the caste system. On the other hand, that the Goddess’s worship was meant for all varṇas and also heretics (pāṣaṇḍas), Tantric physicians (gāruḍikas) and Buddhists, is still registered by the slightly later Devīpurāṇa, in which an ecumenical devotee-base, including even women, is envisaged in such verses as Devīpurāṇa 91.1³⁶ and 35.17 cd³⁷ (on the right of women to worship and the inherence of the Goddess in girls), 22.24 ab³⁸ (on the worship of the Goddess by all varṇas including śūdras) and 88.1–3³⁹ (on the Goddess’s worship by heretics, Tantric physicians, Buddhists and those engaged in other faiths). After the fourteenth century, and the rise of the Rajput lineages in Rajasthan, she became, nonetheless, even more strictly connected with a specifically kṣatriya-ethos despite being, in practice, a non-sectarian deity in the earlier classical period.⁴⁰

One manner in which the Navarātra negotiated the boundary between different religious affiliations is through allowing optionality: nearly all manuals include options allowing the substitution of animals with vegetables, and if one was averse to cutting a human head, about which the Eastern manuals are quite direct, one could just as well manage with a pumpkin. The template was fluid and could be adapted to differing tastes and needs.