re-newing the ancient: the Kāśīkhaṇḍa and Śaiva Vārāṇasī

The present essay reassesses the central narratives of that renowned Purāṇic ‘glorification’ (māhātmya) of Vārāṇasī, the Kāśīkhaṇḍa . In retelling the ancient stories pertaining to Śaiva Vārāṇasī, the Kāśīkhaṇḍa embeds itself within the authoritative tradition of Vārāṇasī māhātmyas, even while effecting an ambitious literary project: a radical reconfiguration of the Śaiva landscape of the city. This reconfiguration would seek to legitimize new Śaiva forms—most prominently, an imperial temple dedicated to Viśveśvara—while reconciling them with Vārāṇasī’s existing Pāśupata infrastructure. Belying facile characterizations of Purāṇa as mere ‘myth’, the Kāśīkhaṇḍa composers took care in ensuring that the many, interwoven strands of its grand narrative of Vārāṇasī’s past were purposefully linked to ideological concerns of the present. A close reading of the Kāśīkhaṇḍa’ s narrative strategies provokes a reevaluation of current scholarly understandings of Vārāṇasī history that view texts as imperfectly reflecting historical realities, rather than as actively constructing that very history.

in his 3 rd century Bce Nirukta, Yāska, the great authority on classical semantic analysis, provides the above-quoted phrase, in which the term purāṇa is analyzed according to its supposed component parts (nirvacana). We might translate this as follows: 'Why is it called Purāṇa? It is because the ancient (purā) becomes new (navaṃ)'. While it seems unlikely that the 'Purāṇa' of Yāska's day would have closely resembled the sprawling web of textual profusion that began to flourish under the title of Purāṇa some centuries later, the mode of oral and/or literary presentation to which Yāska here refers may also underlie the constitution of the class of literature known to us as the 'classical' Purāṇic corpus. In what ways, then, might a Purāṇa be said to 'renew the ancient'?
Though Yāska's insightful gloss is not necessarily supported by a purely linguistic analysis, 1 it certainly calls to mind the oft-cited fluidity of the Purāṇic textual tradition as scholars have encountered it. Any given Purāṇa text can be shown to feature a great deal of interpolation, borrowing, and textual manipulation on the part of various hands, making most research that aims to privilege presumed 'authentic' Purāṇa versions a study in frustration. 2 indeed, there is ample evidence to show that the Purāṇas themselves were frequently explicit about their being subject to change and adaptation, even while affirming their eternality and their origins in divine revelation. 3 It may well be that custodians of the Purāṇa were expected to adapt it to their particular local and temporal contexts, thus making particular strands of the loom of 'ancient lore' new and relevant for particular audiences. this is most apparent in the sub-genre of the sthalapurāṇa: 'Place-Purāṇas' that comprise accounts of traditional lore pertaining to particular tīrthas.
Whether or not this 'updating' of Purāṇic lore was expected by the tradition, Purāṇas were indeed updated, frequently and liberally. Understanding this process as an integral function of the Purāṇas may alter our understanding of this literary genre. rather than viewing them as imperfectly preserving elusive kernels of historical fact, almost impossibly obscured by the accretions of mythological embellishment, we might rather understand the Purāṇas as actively articulating particular versions of the past with specific intent. From this perspective, 'mythical' Purāṇic narratives are not merely accidental fantasies obscuring an imagined core of 'real' history, but rather the narratives themselves are dynamic and purposeful interventions, the intentions of which a discerning examination may occasionally reveal.
This essay will briefly trace one Purāṇic narrative tradition in order to catch one such glimpse of how a Purāṇa 'renews the ancient'. The focus here will be the deployment of the figure of king Divodāsa, ancient ruler of Vārāṇasī, in that most illustrious of Purāṇic 'glorifications' (māhātmya) of Vārāṇasī: the Kāśīkhaṇḍa (Tripāṭhī 1991(Tripāṭhī -1998. though the Kāśīkhaṇḍa, which i date to the latter part of the 11 th century, 4 is historically the most widely distributed and well known of Vārāṇasī sthalapurāṇas (Purāṇa texts focusing on a specific holy place), it is by no means the only one. The Kāśīkhaṇḍa's glorification of the kṣetra (sacred 'field') of Vārāṇasī builds upon a long tradition of such texts, dating back at least to the 6 th century Skandapurāṇa. 5 given that the Divodāsa narrative is a central focus in both texts and that most of the other narratives of the Skandapurāṇa Vārāṇāsī māhātmya find a place in the later text, it seems evident that the Skandapurāṇa serves as the primary māhātmya model for the Kāśīkhaṇḍa. 6 Alhough māhātmya texts are known for their repetitive and formulaic praises of their respective subjects, the details of their individual presentations can vary dramatically. indeed, the Kāśīkhaṇḍa presents a vision of Vārāṇasī that is quite distinct from those of its sthalapurāṇa predecessors: specifically, a Śaiva worldview that may well be attributable to an influential lineage of Śaiva Siddhānta proponents known as the Mattamayūras, who I consider to be the likely composers of the text. this claim diverges considerably from common scholarly views of the Kāśīkhaṇḍa, and hinges on the text's presentation of the famed Viśveśvara liṅga, the story of which is told as the culmination of the narrative of Divodāsa, the ancient king of Vārāṇasī. What follows is an examination of this narrative and a discussion of its implications for the study of Vārāṇasī history.

Vārāṇasī's original ruler? Divodāsa before the Kāśīkhaṇḍa
The story of Divodāsa constitutes the opening narrative of the older Skandapurāṇa māhātmya and relates what I understand to be the underlying theme of the text: Śiva's arrival in Vārāṇasī, where he becomes the foundation of a newly articulated orthoprax Śaiva culture that the text is clearly at pains to articulate. For the later Kāśīkhaṇḍa, 5 the older Skandapurāṇa text bears no relationship to the later series of texts, including prominently the Kāśīkhaṇḍa itself, that identify themselves as 'sections' (khaṇḍa) of the 'Skandapurāṇa'. A critical edition of this older Skanda, based on manuscripts dating to the eighth century ce, is in progress under a team of researchers led by hans Bakker and harunaga isaacson. At the time of this writing, three volumes of the Skanda series have appeared, including a volume (2A) dedicated to the Vārāṇasī māhātmya chapters. See Adriaensen, Bakker, Isaacson 1998;Bakker, Isaacson 2004;Bisschop 2006. In addition, Bakker has published extensively on the significance of the Skandapurāṇa for studies of Vārāṇasī history, and my own analyses draw heavily from Bakker's work, even while diverging from it in some crucial ways. Bakker's most detailed treatment of the subject occurs in the above-cited volume 2A of the Skandapurāṇa series. Also see, most recently, Bakker 2006. 6 Other important Vārāṇasī māhātmyas were composed in the centuries-long interim between the Skandapurāṇa and the Kāśīkhaṇḍa. these include the Matsyapurāṇa (8 th century), a text which almost wholly replicates the Skanda account of Vārāṇasī. Though I exclusively refer to the Skanda here, the Matsya could have just as easily served as the model for the Kāśīkhaṇḍa. Also preceding the Kāśīkhaṇḍa account is the Kūrmapurāṇa (8 th -9 th century?), as well as a Vārāṇasīmāhātmya attributed to an unpublished 'Pāṭālakhaṇḍa' of the Padmapurāṇa. this latter text is well-represented in manuscript form and awaits critical study. On the specifics of the Matsya's indebtedness to the Skanda, see Bakker, Isaacson 2004, 8-9. the importance of the Divodāsa can hardly be missed. Far more elaborate than any previous account, the Divodāsa story is undoubtedly the most central of Kāśīkhaṇḍa narratives, garnering chapter after chapter of description. it also helps form the basis for a central literary trope of that text: the passionate longing that arises from being separated from Vārāṇasī, as from a lover. 7 Because of Divodāsa, the entire pantheon of great and minor gods, including even the supreme Śiva himself, must admire Kāśī from afar, constantly plotting a means to return.
Yet the figure of Divodāsa was not created anew by the Vārāṇasī sthalapurāṇa tradition; it was an inherited narrative, even for the Skandapurāṇa. the early sources for Divodāsa include the Mahābhārata and the so-called Purāṇapañcalakṣaṇa, which together represent some of the earliest available textual layers for the Purāṇas. 8 there are several variants of the story, and it is difficult to identify a consistent narrative, much less to reconstruct a detailed genealogy of its transformations from text to text. hans Bakker, however, has identified some important themes which remain more or less consistent. To summarize his conclusions, Divodāsa is an ancient king of the Kāśī territory (janapada), ruling from the capital of the region, Vārāṇasī. Most accounts focus on the fact that he vies for control of the city with a clan known as the haihayas. the most remarkable and consistently depicted aspect of this rivalry is the fact that Divodāsa is ousted from the city by his enemies, and he is sometimes said to have built a 'second Vārāṇasī' some distance to the north. After one or more generations, the descendants of Divodāsa later recapture the city and repopulate it. In the interim, the city is sometimes said to have been occupied by rākṣasas, under one Kṣemaka. In some versions, Divodāsa's ignominious eviction is attributed to a curse, rather than to the superior military might of his enemies (Bakker, Isaacson 2004, 188-90).
Adding to these observations, it might also be pointed out that there is a crucial 'cultural' dimension running through these early accounts. Divodāsa and his clan are often explicitly identified with orthoprax varṇa culture and the royal sponsorship of the Vedic sacrifice, in contrast with their barbaric or even demonic enemies who 7 Besides its relevance in the Divodāsa narrative and countless poignant references throughout the māhātmya, this familiar Kāvya trope of 'passionate love in separation', vipralambhaśr̥ṅgāra, is most clearly seen in the framing narrative of the text, wherein the sage Agastya leaves Vārāṇasī for the south in order to subdue the Vindhya mountain, who has arrogantly swelled himself so that his increased height has impeded the progress of the sun. the mountain humbly bows down upon seeing the sage approach, and Agastya commands him to remain that way until he, Agastya, returns. Because of this, Agastya can never return to Vārāṇasī, as the journey would require passage through the Vindhyas, and he must thus wander the southern lands forever. eventually, Agastya encounters the god skanda, who narrates the māhātmya of Vārāṇasī with the specific intent of cooling Agastya's fever arising from Kāśī-separation. 8 For textual references to some of these stories, along with a valiant but somewhat problematic attempt to reconstruct the relevant kingly lineages, see Pargiter 1997, 153-6. seem to be outside of the pale of varṇa society. 9 For the early accounts which form the basis of the Purāṇas, then, Divodāsa was not just an ancient king of the Kāśīs, but perhaps the original Vedic king of the region, one credited with introducing-with considerable difficulty, it would seem-the foundational varṇadharma to the region. Though I would hesitate to draw any specific historical conclusions far beyond this, it seems not at all unreasonable to read the early stories of Divodāsa and Vārāṇasī as narratives of the struggle to establish a self-consciously Vedic kingship and culture in the region. given that the centrality of varṇāśramadharma is one of the few truly unifying features of the diverse Purāṇic corpus, this dimension of the narrative is not to be overlooked. 10 For the present study, the most crucial modification to the Divodāsa tale occurs with the dramatic introduction of the god Śiva into the narrative. A few early variants attribute the cursing of Divodāsa to a figure named Nikumbha, who is sometimes identified as an attendant (gaṇa) of Śiva. The Skandapurāṇa māhātmya apparently bases itself on this nikumbha version, and elaborates the details and circumstances of this curse.
in the Skandapurāṇa, Śiva dispatches Nikumbha only after being pressured by his wife, Pārvatī, to find a suitable, civilized place to live. The lifestyle of the divine couple in the remote mountain wilderness, where they constantly dally playfully and cavort with Śiva's raucous gaṇas, has earned Pārvatī the scorn of her parents. Impelled by his wife, Śiva selects the glorious city of Vārāṇasī for their new home, but considers it inappropriate to evict Divodāsa, Vārāṇasī's righteous ruler, without just cause. Nikumbha is sent to trick the ever-righteous Divodāsa into making an error of judgment that would merit his expulsion from the city. nikumbha then manifests himself in a dream of a lowly barber of the town, instructing him to build a shrine in honour of the gaṇa near the palace of king Divodāsa, promising to make the barber rich and prosperous if he should comply. soon the shrine of nikumbha becomes known among the people for being especially powerful, with nikumbha unstintingly granting the desires of all who pay him homage. This prompts Divodāsa's childless 9 This feature of the early narratives becomes more significant when taking into account the fact that the earliest references to Vārāṇasī and the Kāśī region in the Vedic literature frequently represent it as a place that had 'lost' the Vedic yajña, or was otherwise an 'uncivilized' region on the frontier of the Kuru-Pañcāla-centered Vedic orthopraxis. This was true despite the fact that the city was one of the most important commercial centers, especially according to the testimony of early Buddhist literature. the most comprehensive account of the evidence for this is collected in Motichandra 1985, 1-4, 16-19. For a further discussion of the implications of Vārāṇasī's dubious reputation in Vedic literature, see Smith 2007, 31-47. 10 indeed, some scholars-correctly, i think-view the incorporation of local traditions of practice into the greater scheme of varṇāśramadharma-based 'Vedic' ideology as perhaps the most fundamental 'process' of Purāṇic literature as a whole. See especially Chakrabarti 2001. Another provocative articulation of this thesis is found in nath 2001. chief queen to repair to the shrine herself, seeking the boon of a son and heir to the throne. Nikumbha, however, flatly refuses to grant the queen her request. This sends king Divodāsa into a rage, which culminates in his violently dismantling the shrine and setting fire to the sacred image.
Finding his opportunity, Nikumbha appears before Divodāsa and chastises him, reminding him that the gods are free to dispense favours at their own discretion and are not subject to any human, king or otherwise. Because of this offense to the gods, Nikumbha curses Divodāsa, mandating that the king would have to abandon his beloved city, and that it would become unpopulated for 1000 years. 11 It is noteworthy that the uncivilized human enemies that were Divodāsa's rivals in earlier accounts disappear from this Skandapurāṇa version. In fact, when Divodāsa vacates the city it is not rākṣasas or barbarian tribes which move in, but the city is said to be entirely devoid of human life. Only Śiva and his motley entourage of gaṇas are present, the only other inhabitants being wild animals. 12 in the Skandapurāṇa, Divodāsa's struggle for supremacy in Vārāṇasī is not with the uncivilized haihayas, but in fact with Śiva himself. This potentially uncomfortable truth, which might imply that Śiva is an enemy of the Vedic orthopraxis that Divodāsa represents, is handled with great finesse in the māhātmya, which seeks to downplay, if not erase, Śiva's extra-Vedic or adharmic origins. The text makes it abundantly clear that Śiva is not outside or inimical to the Vedic varṇāśrama culture, but is in fact its true master and centrepiece. 13 Once Divodāsa, the original ruler of Vārāṇasī, is evicted, Śiva, Pārvatī, and the retinue of gaṇas occupy the city and settle in, embedding themselves there permanently, where they are to remain in iconic form even after the duration of the curse expires and the great-grandson of Divodāsa-Alarka by name-returns to repopulate the city. 14 this is how, the māhātmya explains, Śiva and his retinue came to abide in the scores of liṅgas and other divine images peppering the landscape of Vārāṇasī.
the most central and powerful of these liṅga shrines, the text describes in copious detail, is Avimukteśvara, where Śiva abides forever in his purest form. Indeed, the 11 The above paragraphs briefly summarize the events of adhyāya 26 of the Skandapurāṇa. the Sanskrit text is available in Bakker, Isaacson 2004, 87-97. 12 atha sā tena śāpena purī vārāṇasī tadā | śunyā samabhavat kṣipraṃ viśuddhā mr̥gasevitā || (26.63) Bakker provides interesting commentary on the significance of mr̥ga in this verse: see Bakker, Isaacson 2004, 192-93. 13 For further elaboration, see Smith 2007, 94-137. 14 alarkaś ca purīm etāṃ matprasādād avāpsyati | sa caināṃ pūrvavat kr̥tvā caturvarṇasamākulām || (30.64) liṅga, and from this the holy 'field' (kṣetra) itself, is said to be called avimukta because it is 'never abandoned' (na mukta) by the supreme lord. 15 Stepping away from the text, this Avimukteśvara liṅga also has great significance for the history of Vārāṇasī and, perhaps, for the early history of Śaivism itself. historians have amassed evidence of an emerging cult of Śaivism in Vārāṇasī beginning not before the fourth century CE, with Avimukteśvara being the most frequently mentioned of Śaiva sites. 16 It seems equally apparent that this early Śaiva presence in Vārāṇasī was of a distinctly Pāśupata character: indeed, the Pāśupatas are the first example of a coherent and distinctive brand of transregional Śaiva praxis and theology. 17 At least by the time of the 6 th century Skandapurāṇa, Vārāṇasī had become the most important centre in a swiftly developing Pāśupata Śaiva network, and within holy Vārāṇasī, the matchless liṅga of Avimukta was supreme. 18

Re-placing the primordial: Śiva and Divodāsa in the Kāśīkhaṇḍa
the Kāśīkhaṇḍa is a māhātmya much different in character than its Skandapurāṇa predecessor. By far the most detailed and best-known text of Purāṇic māhātmyas of Vārāṇasī, the Kāśīkhaṇḍa achieves a comprehensiveness, logical consistency, and linguistic sophistication matched by very few texts in all of classical Purāṇic literature. Weighing in at a rather astonishing 100 ādhyāyas, the Kāśīkhaṇḍa is a massive and impressively enduring composition that seems to have set the standard for Purāṇic māhātmyas. 19 By the 15 th century, the text came to be so widely distributed as to all 15 kadācin na mayā muktam avimuktaṃ tato'bhavat || (29.56cd) 16 For the most thorough and authoritative discussion to date, see Bakker, Isaacson 2004, 21-7, Also see Agrawala, Agrawala, 1984 Despite this, the history of this group is lamentably understudied, with the focus of scholarly inquiries being devoted almost exclusively to the Pāśupatasūtra, Gaṇakārikā, and their attendant commentaries, these being the only texts explicitly labeled as 'Pāśupata' works. There are, however, a vast number of 'Śaiva' texts which were likely authored by Pāśupata adherents. Among these are the early strata of Śaiva Purāṇas themselves. The best study of the Pāśupatas and scholarship thereupon remains lorenzen 1991. 18 This, however, must be qualified by the fact that the Pāśupatas, due to their proliferation, had developed several (perhaps competing) regional identities, each of which would naturally imagine the transregional Pāśupata landscape in different ways. This, I believe, is the significance of Bischopp's observations concerning the distinctions between the recensions of the Skanda. in short, he observes that the Revā and Ambikā recensions of the Skandapurāṇa seem to afford a slightly less prominent place to Vārāṇasī in their modified versions of the text. These precise issues are dealt with in copious detail in Bisschop 2006. For Vārāṇasī in particular, see, for example, ibid., 9. Still, it must be acknowledged that even when other sites are included in the various Skanda recensions, Vārāṇasī retains a very high status. As Bisschop affirms, '[t]here can be no doubt that Avimukta, the sacred site of Vārāṇasī, was considered to be the most holy sanctuary on earth by the composer(s) of the Skandapurāṇa'. see ibid., 18. 19 there is a great need for more critical study of the Kāśīkhaṇḍa manuscript tradition. my own but eclipse the previous māhātmyas of Vārāṇasī. 20 the Kāśīkhaṇḍa was translated, summarized, and commented upon with increasing frequency in the centuries after its emergence, especially from the 15 th century onwards. 21 The story of Divodāsa begins in adhyāya 39 and does not conclude until the description of the triumphant arrival of Śiva-as-Viśveśvara, narrated in the closing chapters of this voluminous text. Even given that it is (in typical Purāṇic style) interspersed with frequent and substantial interludes, the story is told in far greater detail than it is in any earlier account. Besides being much larger, the Kāśīkhaṇḍa version of the Divodāsa story departs from the Skandapurāṇa in other ways; indeed, the respective internal logic of the two texts mandates that this would be so. For one, the understanding of Vārāṇasī espoused in the Kāśīkhaṇḍa as a whole has taken the city beyond the level of being 'merely' the greatest of Śaiva tīrthas, as it is depicted in the Skanda. For the Kāśīkhaṇḍa, Kāśī is not only the greatest of tīrthas, but it is in fact an intrinsic aspect of the universal creation.
the text explains that the primal kṣetra of Avimukta is created before the cosmos itself, arising from the soles of the feet of Śiva and Pārvatī in their cosmic roles as puruṣa, the absolute, unmanifest principle, and prakr̥ti, the primordial womb of the variegated universe. the two-puruṣa and prakr̥ti-'never leave' that kṣetra, the primal ground upon which both forever stand, and so it is called avimukta (26.18-27). 22 Avimukta is thus understood as the ground for the divine coupling which gives rise to the manifest universe and all its living beings.
given this cosmic understanding of the Avimukta kṣetra, there is no question of Divodāsa being the 'original' king of Vārāṇasī whose rule would somehow complicate Śiva's presence. Vārāṇasī, as the primordial kṣetra which serves as a ground for the sport for Śiva and Pārvatī as puruṣa and prakr̥ti, belongs to Śiva, primordially and eternally. in order to accommodate this considerably augmented, cosmic understanding of the significance of the kṣetra, the story of Divodāsa is accordingly initial explorations have confirmed that the text is well represented in manuscript collections all across india, and that the 100 adhyāya format of the published version seems to be standard. it would be interesting to find other variants of the text, or to determine whether the 100-adhyāya form was preceded by an older version. 20 the old Skandapurāṇa with its māhātmya of Vārāṇasī, for example, seems to have been all but forgotten as early as the 13 th century. Adriaensen, Bakker, isaacson 1998, 16. 21 note, for example, the preponderance of Kāśīkhaṇḍa 'summaries' with titles such as Kāśīkhaṇḍasāra, Kāśīkhaṇḍakathāsaṃgraha and Kāśīkhaṇḍacampū. See Minkowski 2002. For details on references to the Kāśīkhaṇḍa in Dharmaśāstra texts, see hazra 1975, 164-65, Adriaensen, Bakker, Isaacson 1998, 15-16. Also see below for a further discussion of the date of the Kāśīkhaṇḍa. 22 unless otherwise noted, all textual citations herein refer to the Kāśīkhaṇḍa edition of Tripāṭhī, 1991Tripāṭhī, -1998 taken to appropriately epic proportions. interestingly, the story is also intertwined with the ancient Avimukteśvara liṅga.
the action takes place far back in mythic time, in a previous manvantara (interval of manu, the name given to the progenitor of the human race after a cycle of dissolution). there was once a great drought which caused the entire earth to become a wilderness, the towns and cities populated by flesh-eaters, the crops having failed everywhere. People had become wild, and chaos and anarchy reigned. Sacrifices were not being performed, and hence the gods, who at that time lived on the earth and not in heavenly realms, were severely weakened because of this (30.26-32). During all this, a great king named Ripuñjaya, born in the noble lineage of Manu, was steadfastly meditating in the ānandakānana (Vārāṇasī, the 'Forest of Bliss'), his senses perfectly controlled. recognizing that only he would be able to restore order to the world by his steady leadership, Brahmā goes to Ripuñjaya and asks him to assume lordship over the earth. Initially, Ripuñjaya humbly demurs, but he finally accepts and later comes to be known as Divodāsa. Divodāsa, however, only accepts rulership of the earth on the significant condition that the gods be banished to the heavenly realms, rather than allow them to continue to live on the earth with men. the explanation given for this is that Divodāsa wishes to rule without a rival (asapatnena rājyena). 23 With this, the text establishes a fundamental rivalry-alongside an uneasy cooperation-between Divodāsa and all the gods.
A few important features are already apparent at this point in the narrative. First, there is the fact that Divodāsa is not a rival only of Śiva and his retinue, as was the case in the more narrowly focused Skanda, but of the entire divine pantheon, with Śiva at its head. This is indicative of the change in the conception of Śiva, who, at the time of the early Purāṇic tradition, seems to have largely still been considered a new and marginal intruder into the brāhmaṇical pantheon, even by the early (Pāśupata) Śaivas themselves. 24 in the Kāśīkhaṇḍa (and indeed, in the later Śaiva Purāṇic tradition as a whole), Śiva has shed most vestiges of his marginality and is firmly established as supreme. The tension between Divodāsa and Śiva can no longer be read as a narrative of Śiva's forced entry into the brāhmaṇical universe-a trope that by now had less relevance-and so the Kāśīkhaṇḍa reinterprets this as a more general cosmic tension between gods and men. in the Kāśīkhaṇḍa, there is no longer a need to justify Śiva's 23 yady ahaṃ pr̥thivīnāthaḥ sarvalokapitāmaha | tadā diviṣado devā divi tiṣṭhantu mā bhuvi || deveṣu divi tiṣṭhatsu mayi tiṣṭhati bhūtale | asapatnena rājyena prajā saukhyam avāpsyati || (39.46-47) 24 The most vivid and recognizable narrative of Śiva's forceful intrusion into the pantheon of elite gods who partake of the sacrifice is, of course, the story of the destruction of Dakṣa's sacrifice. For a discussion and references, see hiltebeitel 1990, 312-35.
Vedic authority, but rather a need to explain why Divodāsa, the righteous, Vedic king should be at all inimical to the gods, who are also, naturally, righteous and Vedic.
The text resolves the tension by expanding the story to a cosmic scale. Divodāsa needed to rule the entire earth (which, as has been shown, is now thought to be literally centred in Vārāṇasī) without distracting the people with other objects of worship: Divodāsa himself is shown to be great enough to fulfil all their needs, so righteous and powerful was he. Moreover, the gods are dependent on sacrifices, and hence they require a stable human rule in order to ensure their regular performance. The effect of all this is that Divodāsa's power is augmented in the Kāśīkhaṇḍa to a spectacular degree. With this, the Kāśīkhaṇḍa solves the problem of the tension between Divodāsa and the gods, and acknowledges the dependency of the latter upon this exemplary human ruler.
This, however, introduces another important logical problem: that of Śiva's supremacy. While the device of Divodāsa's demand that the gods be banished to heaven makes sense with respect to the gods in general, it would be an unthinkable diminution of Śiva's power to depict the Śiva himself as being subject to this demand. As the supreme lord of the universe, Śiva is not dependent on sacrifices or any other human agency; and yet he must leave Vārāṇasī, his primordial abode, and trick his way back in, in order for the central story to unfold. in recasting this essential story in the image of Vārāṇasī that they want to project, the Kāśīkhaṇḍa composers introduce the problem of having to explain why and in what sense Śiva 'abandoned' Kāśī in the first place, allowing Divodāsa to establish himself there to his own exclusion.
the Kāśīkhaṇḍa composers adroitly solve this further paradox. just at the crucial moment when the gods departed for the heavens, leaving Divodāsa alone to rule all humans, the text introduces an interlude, wherein Śiva travels to visit Mount Mandara to grant a boon to this devoted mountain, who has been busily practicing austerities in order to gain Śiva's favour (39.50-53). Śiva expresses his pleasure at Mandara's austere efforts, and, as promised, grants him a boon. Overjoyed at receiving the grace of the lord, mandara makes his request: he wishes that he be made 'equivalent to the Avimukta kṣetra', and that Śiva and his entourage should make their residence upon the devoted mountain's 'head'. 25 || (39.56-59) to Śiva that he, Brahmā, had promised Divodāsa that all the gods would be leaving the earth and allow him to rule without a rival, hastening to add, diplomatically, that this would only be temporary arrangement-after all, the mortal's lifespan, and even the vast lifespan of Brahmā, the creator of the universe, is hardly an eye-blink in the reckoning of the great and eternal Lord Śiva. Then Śiva grants Mandara the boon of being equal to Avimukta, knowing that he would have to stay there for some time, given Brahmā's hastily made arrangement with Divodāsa (39.60-70).
remarkably, this is not the only strategy that the Kāśīkhaṇḍa employs to explain Śiva's apparent absence from Vārāṇasī. In this second narrative explanation, moreover, the old Pāśupata sanctuary, the Avimukteśvara liṅga, is a crucial narrative component.
the māhātmya reveals that, unknown even to Brahmā, Śiva had already established the Avimutkeśvara liṅga, an image of himself, before leaving Kāśī, in order to protect the kṣetra and to grant salvation to devotees in his 'absence'. 26 Śiva, then, has not really left Avimukta in two ways: he has made his temporary abode, mount mandara, 'equivalent to Avimukta', and he had moreover established himself in Vārāṇasī liṅga form before leaving. In effect, this makes Avimukteśvara the original liṅga form of Śiva, installed by Śiva himself. The text emphasizes the importance of this foundational act, which also, mirroring the Skanda, serves as an updated explanation of the significance of the name avimukta: And so, even when the Lord who bears the Pināka bow Went to mt. mandara, this kṣetra was not abandoned (vimukta) by him, in his liṅga form: It is, therefore, known as Avimukta, "Never-Abandoned". long ago, this kṣetra was celebrated as Nandavana, the Forest of Delight, But from that point forward, its name on earth came to be Avimukta. Avimukta became the name of both the kṣetra and the liṅgahaving come to these two, one never again has to reside in a womb. Upon seeing this Avimukteśvara liṅga in the Avimukta kṣetra, One truly becomes vimukta: 'freed' from all the bondage of karma. Despite its resemblance to the Skandapurāṇa quoted earlier, this apparently formulaic glorification of the Avimukteśvara liṅga and the rest of the kṣetra has a much different function in the Kāśīkhaṇḍa. While other liṅgas are set up by divine and human devotees in honor of Śiva, Avimukteśvara is here said to be installed by Śiva himself. This would seem to indicate that the Kāśīkhaṇḍa, in keeping with the previous Vārāṇasī sthalapurāṇa tradition, considers Avimukteśvara to be the supreme liṅga in Vārāṇasī, and indeed, in the universe. The Kāśīkhaṇḍa, however, reserves that distinction for another liṅga, one which, significantly, is all but unknown before the Kāśīkhaṇḍa itself but swiftly becomes the most recognizable and central of Kāśī's innumerable holy sites: the liṅga known as Viśveśvara.

'New' Vārāṇasī:
The literary construction of the Viśveśvara temple the remainder of the Divodāsa narrative illustrates, in copious and meandering narrative detail, the great efforts that the gods, directed by Śiva himself, exert in order to gain re-entrance to Kāśī, having been parted from her because of Brahmā's boon to Divodāsa. The basic structure of the remainder of the text follows, though in much more elaborate form, the model of the Skandapurāṇa. Śiva sends his representatives to Kāśī in order to trick the ever-righteous Divodāsa into making a mistake in dharma, so that he may finally be legitimately forced to leave the city, allowing the gods to once again enter. In this text, however, Śiva does not simply send a single gaṇa to accomplish this purpose: given his augmented status in the Kāśīkhaṇḍa, Śiva now has the entire pantheon of gods, hierarchically ordered from yoginīs and gaṇas all the way up to Brahmā, Gaṇeśa and Viṣṇu, under his direction. And he will need all of this help, as Divodāsa is considerably more powerful and less prone to slippage in this text. the lower orders of divinities will prove unable to shake his righteous rule, causing Śiva to dispatch increasingly powerful gods to achieve his purpose, not succeeding until Gaṇeśa exercises his peerless cleverness, and Viṣṇu then arrives to deal the final blow, at last bringing about Divodāsa's fall. Only then will the path be cleared for Śiva's triumphant return to the city as Viśveśvara, Lord of the Universe.
the lavishly recounted details of the Kāśīkhaṇḍa's presentation of the gods' tireless efforts to unseat the powerful Divodāsa have been elegantly summarized elsewhere (Eck 1999, 148-57). Relevant here is the culmination of the story, in which Śiva's return is described. There, as Viśveśvara is formally welcomed to the city in a grand ceremony, Śiva does not establish another liṅga as in the case of Avimukteśvara, but he installs himself in permanent liṅga form as 'Viśveśvara'.
it is certain, moreover, that this liṅga is established in a 'temple' setting, and this temple housing Śiva's greatest manifestation was appropriately elaborate. Upon the departure of Divodāsa, Nandī, one of Śiva's most trusted gaṇas, informs Śiva that the work for the construction of a great 'palace' (mahāprāsāda) has been completed, and that all of the gods have assembled to welcome him to it. 28 the triumphant arrival of Viśveśvara is described in a way that parallels, rather self-consciously, it would seem, a specific ceremony (mahotsava), commemorating Śiva's 'entrance' (prāveśika) into his 'palace' (prāsāda): that is, the ritual installation of the deity in the temple. 29 this moment appears to be of special significance for the text, which then breaks from the conventional śloka meter into a more elaborate and rousing upajāti, continuing in this meter for several verses which describe the worship ceremony. 30 the assembled divinities worship Viśveśvara in a style which mirrors-or perhaps more accurately, serves as a model for-ceremonial temple pūjā, including the offerings of flowers, jewels, garments, and flower garlands. Also mentioned are the rites of abhiṣeka, the ceremonial bathing of the liṅga in the style of a head-bath for royal coronations, and nīrājana, the concluding lamp offering.
the text also describes various aspects of the layout of this spacious temple compound, including the much-lauded Freedom Pavilion (muktimaṇḍapa): probably an expansive, covered platform located to the south of the inner sanctum (garbhagr̥ha), and for this reason it is also called the southern Pavilion (dakṣiṇamaṇḍapa). 31 this Freedom Pavilion seems to have been a very important and distinctive feature of the temple compound, as the text provides a lengthy sub-narrative explaining an 'ancient' name of the place: the Pavilion of roosters (kukkuṭamaṇḍapa). 32 it seems to have been known as a place of repose and peaceful meditation and is elsewhere recommended as a place of rest after an exhausting pilgrimage circuit (yātrā) around 28 jātā parisamāptiś ca mahāprāsādanirmiteḥ | sajjīkr̥to rathaś cāyaṃ brahmādyā militāḥ surāḥ || (97.294) 29 śr̥ṇu sūta mahābhāga yathā skandena bhāṣitaḥ | mahāmahotsava śambhoḥ pr̥cchate kumbhasaṃbhave || (98.1) sarvaṃ pramuditaṃ cāsīc chambhoḥ prāveśikotsave | (98.7cd) 30 āgatya devadevo'tha muktimaṇḍapam āviśat || (98.20cd) the upajāti verses which immediately follow describe the worship in the muktimaṇḍapa, 98.21-37. 31 Besides the mukti-and śr̥ṅgāra-maṇḍapas, no other pavilions are mentioned in the Kāśīkhaṇḍa description. A later Viśvanātha temple, the plan of which was sketched by James Prinsep, shows a series of four maṇḍapas surrounding the garbhagr̥ha: the names of the pavilions are mukti, śr̥ṅgāra, aiśvarya and jñāna corresponding to the south, west, north and east of the central image. these pavilions, however, were each only 16 feet square and the two pavilions of the temple described by the Kāśīkhaṇḍa were probably considerably larger, though it is certainly possible that this is a case of māhātmya exaggeration. this later temple upon which Prinsep's ground plan is based was dismantled by Aurangzeb in 1669, and the Ālamgīr mosque was erected from its rubble. While it is possible that the later incarnations of the Viśvanātha temple were modeled on the original described in the Kāśīkhaṇḍa, the descriptions in the text are in many places inconsistent with the later plans, making this unlikely. For a wonderfully detailed discussion of the layout, based on the plan of Prinsep, of the temple destroyed by Aurangzeb, see Altekar 1937, 50-8. 32 more on this curious reference below.
Kāśī's innumerable holy sites. 33 Another part of the temple compound features a Decoration Pavilion (śr̥ṅgāramaṇḍapa; raṅgamaṇḍapa); here, Śiva is said to assemble his entourage and look upon Viśveśvara liṅga, recounting the glories of the liṅga, of the nearby Maṇikarṇikā tank, and of the city of Kāśī as a whole (99.3-61). 34 At last, Śiva rises and performs the worship of the Viśveśvara liṅga, and finally merges into the liṅga, while the gods shout in praise and triumph. 35 Śiva then declares, addressing the full pantheon of gods who had come to celebrate his presence: This is Viśveśvara, the Lord of the Universe, Manifested here in fixed form, the grantor of all powers to those endowed with devotion. sometimes seen, and sometimes unseen, O gods, I always reside here in this Forest of Bliss (ānandakānane) as i please, to shed grace upon all devotees. i abide here in liṅga form, granting what is wished for. these liṅgas all around here-Both naturally occurring (svayaṃbhū) and installed (asvayaṃbhū)-All constantly come here to see this liṅga. While there is no doubt that I reside in all liṅgas, this is the supreme image of me in liṅga form. Whoever sees this liṅga with faith and pure vision, O heaven-dwellers, Verily sees me, manifested! 36 33 muktimaṇḍapikāyāṃ ca kṣaṇaṃ yat sthiram āsyate | snātvā gaṅgāmr̥te śuddhe tapa etad ihottamam || (39.8cd-9ab) iti mantraṃ samuccārya kṣaṇaṃ vai muktimaṇḍape | viśramya yāyād bhavanaṃ niṣpāpaḥ puṇyavān naraḥ || (100.97) 34 the name of this second maṇḍapa suggests an area where the deity would be taken out annually for a special decoration ceremony, thus continuing the explicit temple terminology of the text. Discussing the current practice in the Viśveśvara temple, Eck describes this deity's current śr̥ṅgāra ceremony: 'On the eleventh day of the [waxing] fortnight [following Śivarātri], the last half of Phālguna, the yearly decoration day (śr̥ṅgāra) of Viśvanātha takes place. After Śivarātri, this is the biggest festival day of the year at Viśvanātha. On this day, the special silver four-faced cap which sits upon the liṅga is set in place and decorated with all the proper sandalwood paste, leaves, and flowers. On this day, however, rather than sprinkling the liṅga with Ganges water, worshippers sprinkle it with colored red powder. The day is called the "Colorful Eleventh", Raṅgabhārī Ekādaśī, and it anticipates the riot of color throwing which will break out four days later on holī' (Eck 1999, 277 Viśveśvara, an epithet of Śiva used throughout this text and elsewhere, is here revealed to be the name of the supreme liṅga form of Śiva (parā liṅgasvarūpiṇī mūrti). It was Viśveśvara himself who had installed and worshipped Avimukteśvara, before leaving (in a sense) Vārāṇasī and finally returning after Divodāsa is ousted. The composers of this text clearly exalt both Avimukteśa and Viśveśvara, but there is no doubt that, while Avimukteśvara is temporally prior, Viśveśvara is supreme. The Kāśī of this text is clearly not the 'old' Avimukta of the Pāśupatas: the Śaiva vision has been 'updated'.

From Purāṇa to 'Itihāsa': Situating Viśveśvara in Vārāṇasī history
The description of this Viśveśvara temple, along with the positioning of this account at the very conclusion of the vital Divodāsa narrative, strongly suggests that a newly constructed edifice is referenced, and that the Kāśīkhaṇḍa is at great pains to ideologically ground this new temple within the ancient Śaiva landscape of the city. As a testament, perhaps, to the success of this project, Viśveśvara was indeed to become the most renowned liṅga of Vārāṇasī, a distinction which continues to the present day. But how does this 'glorified' account of this māhātmya accord with what we know of Vārāṇasī history?
At first glance, it does not. At the root of the issue is the relationship between Avimukteśvara and Viśveśvara, an issue which has baffled scholars both traditional and modern. Because of the close connection between these two liṅgas, several scholars, both medieval and modern, have considered Avimukteśvara and Viśveśvara to be two names of the same deity. 37 As eck remarks, this is in some sense true, 'for each in its own day has been the pre-eminent shiva liṅga of Vārāṇasī' (Eck 1999, 129). Most modern scholars hold that Avimukteśvara, the ancient Pāśupata shrine, was the most sacred liṅga of the city from the 4 th century until perhaps the 12 th . A short time thereafter, Viśveśvara, a liṅga which is hardly ever mentioned in earlier accounts, rather suddenly emerges as the supreme manifestation of Śiva according 37 Eck notes that Vācaspati Miśra makes this assertion in his Tīrthacintāmaṇi (1460), as does Nārāyaṇa Bhaṭṭa in his Tristhalīsetu (16 th c.). in the modern period, eck quotes ram shankar tripathi, a former mahant of the Viśvanātha temple, as sharing this view; see Eck 1999, 129-32. So confused is the history of the relationship between the two temples that the earliest modern authors make no mention of Avimukteśvara at all, allowing A.S. Altekar, who wrote the first dedicated historical study of Vārāṇasī in 1937, to claim that '[f]or more than the last two thousand years at least Benares has been famous as the place of the temple of Visvanātha'. See Altekar 1937, 44. Reverend M.A. Sherring, the grandfather of Vārāṇasī studies, operates under the same error in omitting Avimukteśvara from his discussion of Bisheśwar (Viśveśvara) as the 'idol-king of Benares'. See Sherring 1975, 47-54. to countless sources, including prominently the Kāśīkhaṇḍa itself. yet the historical causes for this 'regime change' remain mysterious. 38 In sorting out this persistent problem, it is significant to note, in the first place, that the descriptions in the māhātmyas suggest that Viśveśvara of the Kāśīkhaṇḍa occupied the very same site as Avimukteśvara once did: the holy ground near the storied tank of Maṇikarṇikā, standing not far from the banks of the northward-flowing waters of the river Gaṅgā. The above-cited verses praising the muktimaṇḍapa of the Viśveśvara compound repeatedly recommend that a pilgrim sit and rest in this pavilion immediately after bathing in the Maṇikarṇika tank, also called the Cakrapuṣkariṇī, the pond supposedly excavated by Viṣṇu's discus and filled with the sweat of his exertion, as a show of his devotion to Lord Śiva. 39 Bakker shows that, at a relatively early point in the history of Vārāṇasī, the cremation ground (śmaśāna) came to be adjacent to Maṇikarṇikā tank, as is famously the case today at Maṇikarṇikā Ghāṭ. This śmaśāna, moreover, was originally known as Avimuktaka, 'which seems to imply that it was situated in the vicinity of the Avimukteśvara temple' (Bakker, Isaacson 2004, 46). more evidence for this is found in the fact that the Kāśīkhaṇḍa provides a narrative to explain an 'ancient' name for the muktimaṇḍapa of the temple compound: the text explains that in the Dvāpara age, the site was called the kukkuṭamaṇḍapa, or Pavilion of roosters. 40 this seemingly incidental detail is explained by a curiously elaborate story of how the pavilion came to acquire this strange epithet, or rather would come to acquire it, since the story is narrated by Śiva upon his return from Mandara, as a prophecy of 'future' events. To briefly summarize this rather meandering story told by Śiva, a corrupt Brahmin named Mahānanda came to reside in Kāśī in the Dvāpara age, and he once accepted a lucrative gift from a rich, low-caste Cāṇḍāla pilgrim. Excommunicated and censured by the Brahmin community, he left Kāśī with his family and was assailed by bandits, who robbed and killed him along with his family. Fortunately, the Mahānanda family remembered their beloved Kāśī at the time of their death and were all reborn as chickens in a land called Kīkaṭa. From there, they followed a group of pilgrims back to Vārāṇasī, and their descendants have remained there in the muktimaṇḍapa ever since, worshipping Śiva and earning the adoration of the pilgrims who witness their piety. Because of this, the text explains, people refer to the Freedom Pavilion as the 'Pavilion of Roosters ' (98.40-89 || (98.38) there is at face value not much remarkable to this story other than its odd specificity, but it seems more than coincidence that roosters play a prominent role in an older Purāṇic story concerning the temple of Avimukteśvara. Occurring in the thus-far untraceable Liṅgapurāṇa quoted by the 12 th century Lakṣmīdhara in his Tīrthavivecanakāṇḍa (which will be discussed further below), 41 this older story also apparently serves as a narrative explanation for the presence of roosters at Śiva's central shrine. indeed, the story concludes in a manner quite similar to the Kāśīkhaṇḍa tale, finally affirming that 'even now, roosters still constantly frequent that place, and are watched and honoured by those of good intentions '. 42 In this older Purāṇa, however, the place described is not the muktimaṇḍapa of Viśveśvara, but the immediate vicinity of the Avimukteśvara liṅga. this rather bizarre coincidence certainly seems to suggest that these two rooster-infested shrines must refer to the very same temple compound, at different historical moments. It seems that Viśveśvara came to replace Avimukteśvara not only in his assumption of the unofficial title of reigning liṅga of Kāśī, but he in fact literally occupied Avimukteśvara's very 'throne'-and along with it, it seems, he inherited a crowing coterie of avian devotees who were living in close proximity.
In attempting to explain the dramatic shift of emphasis from Avimukteśvara to Viśveśvara around the 12 th century, scholars tend to point to that persistent bugbear of studies of medieval south Asian history: the destruction of temples in islamic invasions. A prevailing notion holds that Viśveśvara is a sort of reincarnation of Avimukteśvara, rebuilt and rechristened after 1194, when the armies of Quṭbuddīn Aibak, the general of Maḥmūd Ghūri and eventual founder of the Delhi Sultanate, attacked the city. 43 Bakker affirms that 'most scholars … have dated the rise of 41 hans Bakker is of the opinion that the unique version of the Liṅgapurāṇa that Lakṣmīdhara quotes 'may have been written during Lakṣmīdhara's lifetime and possibly at his instigation' (Bakker, Isaacson 2004, 71). I am rather inclined to believe that, given its heavily Pāśupata character and the high degree of specificity provided with regard to the shrines of Vārāṇāsī, this text, which is quoted extensively in Lakṣmīdhara's work, was a previously existing Pāśupata work: one that perhaps never circulated widely beyond Vārāṇasī. See Smith 2007, 253, n. 411. 42 kukkuṭāś cāpi deveśi tasmin sthāne sthitāḥ sadā | adyāpi tatra dr̥śyante pūjyamānāḥ śubhārthibhiḥ || (Aiyangar 1942, 109) 43 strictly speaking, eck, following Kubernath sukul and elaborating on the observations of Sherring, argues that an older Viśveśvara originally stood some distance away from Avimukteśvara, atop a nearby hill at the site of the temple now known as the 'Original Viśveśvara' (Ādi-Viśveśvara). When both temples were destroyed in 1194, the new liṅga at the site of the old Avimukteśvara temple was called Viśveśvara, because of the popularity of the former Viśveśvara temple among the 'commonfolk', this latter distinction an explanation for the fact that hardly any references to Viśveśvara exist before the Kāśīkhaṇḍa itself (Eck 1999, 129-34). Compare Sukul 1974, 176-7; see also Sherring 1975, 54-5. Altekar, on the other hand, rightfully questions Sherring's easy reading of the 'Ādi' Viśveśvara temple as being the 'original' temple. see Altekar 1937, 44. in my view, the ādi-, historically speaking, is indication not of priority but just the opposite: it was probably named as such precisely be-Viśveśvara/Viśvanātha to the position of presiding deity of Vārāṇasī to the century after the sack of the town' by Aibak (Bakker, isaacson 2004, 72).
But these speculations about these important monuments are themselves on very 'shaky ground'. As Bakker himself shows, though without extensive comment on the implications of this important evidence, there are references to Kāśī's Viśvanātha or Viśveśvara temple which definitively precede the 1194 attack of Aibak by some decades (ibid., 72-5, and see below). There has not yet been an evaluation of exactly what this means for the study of Vārāṇasī history, but in placing this evidence alongside the sthalapurāṇa tradition, a fascinating picture of shifting patterns of Śaiva patronage in Vārāṇasī begins to emerge. It is significant to note, for one, that these early references to Viśveśvara of Kāśī are invariably of explicit Śaiva Siddhānta orientation, an issue that will be discussed in some detail below.
in any case, scholars are also nearly unanimous in considering the Kāśīkhaṇḍa to be a text that dates to a period after this transition from Avimukteśvara to Viśveśvara. To a large extent, it is the very fact that Viśveśvara is so prominently foregrounded in the text that the Kāśīkhaṇḍa is assumed to have been composed in a period after which Viśveśvara had already assumed the status of prime liṅga in Vārāṇasī. As we have seen, however, a close reading of the text itself strongly suggests not that the Kāśīkhaṇḍa records this as fact, but in fact precisely intends to effect this reality with its creative re-imagining of the past. This is cause enough to briefly reassess what we know of the origins of this text. most of the early mentions of a Kāśīkhaṇḍa text by name are found in Dharmaśāstra digests (nibandhas) including what is probably the earliest reference, the Caturvargacintāmaṇi of hemādri (13 th century) (hazra 1975, 163-4). The quotes that hemādri ascribes to this 'Skānda-Kāśīkhaṇḍa' are not, however, found in the version available to us today (Adriaensen, Bakker, isaacson 1998, 15). the earliest references to verses that do correspond to the extant version occur in the Madanaparijāta of Viśveśvarabhaṭṭa (late 14 th century), where they are ascribed to a 'Skandapurāṇa' (ibid). By the 15 th century, references to the Kāśīkhaṇḍa are legion.
While the great Purāṇa authority, R.C. hazra, asserts, based on this evidence, that the Kāśīkhaṇḍa 'has chapters … which are older than 1300' CE (hazra 1975, 165), Vārāṇasī scholars such as Kubernath Sukul, Diana Eck and, more recently, hans Bakker tend to date the current version of the Kāśīkhaṇḍa to between the late 13 th and mid 14 th centuries (Eck 1999, 47;Sukul 1977, 287-9;Adriaensen, Bakker, Isaacson 1998, 15). As mentioned, this period has gained favour due in part to the fact that cause it was a newer construction seeking to assert an ancient pedigree. Sukul finds this pattern with the Vr̥ddha-Kāleśvara shrine, thinking it to be a reconstruction of a previous temple to Kāleśvara. See Sukul 1977, 350-1. For further discussion, see Smith 2007 Viśveśvara is mentioned so prominently in the text. But it depends even more on one important apparent piece of negative evidence: the fact that it is not quoted in the late 12 th century Dharmaśāstra compendium, the Kr̥tyakalpataru, which features in its Tīrthavivecanakāṇḍa an extended section which compiles Purāṇic quotes on Vārāṇasī. The author of this text, Bhaṭṭa Lakṣmīdhara, was a minister of the ruler Govindacandra (r. 1114-1154), the greatest of the kings of the Gāhaḍavāla dynasty. Discussions of Vārāṇasī history frequently highlight this relatively short-lived but profoundly influential dynasty, which ruled from Vārāṇasī itself throughout much of the 12 th century. This Gāhaḍavāla period marks the first time in the long history of Vārāṇasī that the city would serve as a capital of a transregional dynasty. The Gāhaḍavāla capital was shifted to Vārāṇasī from Kānyakubja, the ancient capital of northern India further up the Gaṅgā, the shift partially motivated by increased pressure from turko-Afghan armies, which had already conquered the sindh region and had been making incursions further into the subcontinent, thus threatening the Indian kings of the northwest. When Quṭbuddīn Aibak, fresh off similarly destructive victories at Kanauj and Ayodhyā, sacked Vārāṇasī in 1194, he not only destroyed temples, but effectively ended the period of Gāhaḍavāla ascendancy.
had the Kāśīkhaṇḍa been in existence during Lakṣmīdhara's time, it is reasonable to assume that Lakṣmīdhara, as the rājaguru of Govindacandra based in Vārāṇasī itself, would have included substantial quotes from the text in his Tīrthavivecanakāṇḍa. Indeed, Lakṣmīdhara devotes more attention to Vārāṇasī than to any other tīrtha, and cites an impressive range of Purāṇa texts. The fact that not even a single verse of the Kāśīkhaṇḍa appears is surely significant. 44 the 12 th century flourit of Lakṣmīdhara and govindacandra, then, appears to serve as a reasonable terminus post quem for the Kāśīkhaṇḍa.
this, however, is far from certain. even if we must put aside, for the present, the tantalizing but as yet uncorroborated reference of hazra, who claims to have seen a reference to the existence of a Kāśīkhaṇḍa manuscript dated 933 Śaka (1011 CE), 45 there remain reasons to question the 13 th -14 th century date for the Kāśīkhaṇḍa. For one, the abundance of historical information from the Gāhaḍavāla period has tended to obscure the importance of the immediate predecessors of the Gāhaḍavālas in Vārāṇasī, the Kalacuri kings of Ḍāhala, the Baghelkhand region of eastern Madhya Pradesh. The Kalacuri empire flourished under the feared conqueror Karṇadeva, 44 For a discussion of the intertextuality between the Tīrthavivecanakāṇḍa and the Skandapurāṇa tradition, see Bisschop 2002. 45 hazra cites the Bengali Viśvakośa, a massive encyclopedia project in 22 volumes, compiled under the direction of Nagendranath Basu between 1886 and 1911; hazra says that the Viśvakośa office itself held the manuscript. I have not yet been able to locate this manuscript, nor yet even the text in which it is supposedly mentioned. see hazra 1975, 165n. who ruled from 1041 to 1072, dominating his rivals while annexing huge territories, including the two most important tīrtha cities of the Gangetic plains, Vārāṇasī and Prayāga (Allahabad). 46 Kalacuri inscriptions, beginning with those of Karṇadeva's son, Yaśaḥkarṇa (1073-1120), make reference to Karṇa as having built an imperial temple in Vārāṇasī. Kalacuri praśastis beginning with those of Yaśaḥkarṇa, who was to eventually cede the Vārāṇasī region to Candradeva, the first Gāhaḍavāla king of note, refer to Karṇa's temple as the Karṇameru; with -meru referring symbolically to the mythical meru world-mountain, the axis-mundi, and also to a particular type of imperial temple construction (Bakker, Isaacson 2004, 65-6). 47 Could this Karṇameru have been the original Viśveśvara temple referred to in the Kāśīkhaṇḍa?
it is, in fact, a possibility. no record of this construction has ever been found, but it may have been right at the heart of the sacred city itself, dismantled and rebuilt several times over the course of almost a millennium. 48 Other references associate Karṇa with imperial temple construction, and also with a Karṇameru. In the early 14 th century, Prabandhacintāmaṇi of the Jain historian Merutuṅga, for example, Karṇa, said to be the ruler of Vārāṇasī, is depicted as competing with Bhoja, the Parāmara ruler of Avantī, to see which of the two would be able to complete the construction of a temple the fastest; the agreement being that the loser of the contest would surrender to the winner. In this fanciful but telling account of Merutuṅga, Karṇa emerges from this temple-building contest victorious (Muni 1933, 50-1). Other passages in the same text specifically reference a Karṇameru: in one place, however, this meru of Karṇa is said to be located not in Vārāṇasī, but in Śrīpattana (Patna) (ibid., 55). It will be recalled that the Viśveśvara shrine in Vārāṇasī had been dismantled and rebuilt more than once by Merutuṅga's time, perhaps explaining the discrepancy. Even if Merutuṅga's account is hardly reliable in its historical details, the association of Karṇa with Vārāṇasī, with temple building in general, and with a Karṇameru in particular, is certainly striking.
What is more, the Kalacuris are known to have had close ties with the Mattamayūras, an important lineage of Śaiva Siddhānta teachers who served as advisers to Kalacuri kings nearly from the origins of the lineage in the 9 th century. Indeed, the rapid spread of Śaiva Siddhānta texts and temples from the 10 th 46 Technically speaking, the region was probably already conquered by Karṇa's father, Gāṅgeyadeva, but as we will see, Karṇa becomes the Kalacuri most associated with Vārāṇasī. For details on the lineage, see Ray 1973, 772-9. 47 For a discussion, see Smith 2007, 281-4. 48 For a detailed, if frequently speculative, history of the various destructions and reconstructions of this shrine, see Altekar 1937, 44-58. century onwards corresponded with the Siddhāntins' ability to develop intimate connections with powerful royal lineages such as the Kalacuris and the colas to the south. Appropriately then, the Siddhānta Āgamas regularly fixate on ritual details pertaining to both temple construction and temple worship on the one hand, and rituals of royal power on the other. the imperial temple is indeed one of the central and distinctive features of Siddhānta discourse.
Tellingly, as briefly mentioned above, the earliest known clear references to the Viśveśvara temple are of explicit Śaiva Siddhānta orientation. A Nepalese inscription of 1143 records the activities of a Śaiva Siddhānta teacher named Rudraśiva, and praises Viśvanātha (=Viśveśvara) as the lord of Kāśī (Bakker, Isaacson 2004, 72-4). 49 Even more explicit is the testimony of the South Indian author Jñānaśiva (12 th century), whose Jñānaratnāvalī contains verses revering both Dabhrasabheśvara Śiva of Cidambaram (another elaborate temple-complex of Śaiva Siddhānta affiliation 50 ), and Viśveśvara of Vārāṇasī (Bakker, Isaacson 2004, 74-5). These references prove that Viśveśvara was definitively in existence before 1194 and that some Siddhāntins, at the very least, considered Viśveśvara to be the presiding deity of Kāśī and one of the most significant sacred sites in the Śaiva world.
And what of Lakṣmīdhara's silence? If the Kāśīkhaṇḍa were a text somehow associated with the Kalacuris and their Śaiva advisers, Lakṣmīdhara, as the chief minister of Govindacandra Gāhaḍavāla, surely would have distanced himself from it. The Kalacuris were bitter rivals whom the Gāhaḍavālas deposed to obtain the Vārāṇasī region, and their close Mattamayūra advisers could not have been held in high favour. indeed, an inscription of govindacandra records his diverting land away from a Kalacuri rājaguru, a Śaiva (Siddhānta) teacher (śaivācarya) named Rudraśivaland which had been given to Rudraśiva by Yaśaḥkarṇa Kalacuri himself-and granting it to a Brahmin hand-picked by govindacandra. 51 i strongly suspect that the teacher disenfranchised by this grant is the same Rudraśiva that is mentioned in the 1143 Nepalese inscription cited earlier. Significantly, this later inscription says that Rudraśiva left for Nepāla when Kāśī, the divine city, was 'enslaved', (dāsīkr̥tasurapurī) 49 see below for a further discussion of this interesting document. 50 On the history and worship rituals of this famous temple, see younger 1995. 51 this 1120 ce grant records govindacandra transferring the ownership of a village named Karaṇḍa to one Ṭhakkura Vasiṣṭha Śarman and his future descendants. See hall 1862, 111-28. The institutional religious affiliations of the Gāhaḍavālas have been the subject of some debate. On the one hand, their praśastis consistently style the kings of the lineage as paramamāheśvaras, a designation which has often been taken (perhaps along with their ruling from the heavily Śaiva-leaning Vārāṇasī) as indicative of their Śaiva predilections. But their own imperial temple in Vārāṇasī, built by Candradeva, was the original Ādikeśava (Viṣṇu) temple, and it was here that their royal ceremonies meant to display the splendor of their lineage were performed. Other evidence makes it quite clear that they were Vaiṣṇava initiates. For a thorough discussion, see Niyogi 1959, 194-5. in other words, 'conquered'. The general time period would fit with the Gāhaḍavālas' usurping of the region from the Yaśaḥkarṇa Kalacuri. 52 if it is the case that the Kāśīkhaṇḍa was a Mattamayūra/Kalacuri production, it is not at all difficult to imagine that Lakṣmīdhara would have deliberately ignored this text out of loyalty to his own royal patron. 53 One wonders further if his own Kr̥tyakalpataru, and its Tīrthavivecanakāṇḍa in particular, might have been intended specifically to challenge the authority of the Kāśīkhaṇḍa, framing the sacred tīrthas, for one, in a manner that was more in line with his own ideological sensibilities. though the Kr̥tyakalpataru is of another genre entirely, this expansive Dharmaśāstra compendium is similar to the Kāśīkhaṇḍa in that it seems to have been composed to stand as a comprehensive, final authority on all things related to dharma, discourse on tīrthas included.
We have seen that it is hardly true that, as Kubernath sukul asserts, 'when the present version of the Kāśīkhaṇḍa was compiled Avimukteśvara had lost his preeminence altogether and Viśveśvara had become all in all in Varanasi' (Sukul 1974, 175). A careful reading of the Divodāsa narrative concerning the two liṅgas makes it evident that there is something at play beyond a banal recording of historical events. The special care with which the text lauds Avimukteśvara as the Original Liṅga (ādiliṅga), along with the text's remarkable specificity with regard to the 52 Note that here I differ with Bakker's reading of the crucial phrase: kāśīṃ dāsīkr̥tasurapurīṃ (…vihāya). Bakker takes this to mean '[Rudraśiva] left Kāśī, which reduces the city of the gods to subordination', a reading which is certainly possible but in my view somewhat awkward. in any case, the inscription goes on to say that later, 'beckoned, as it were, by Viśvanātha himself', Rudraśiva returned briefly to Kāśī and sponsored the reconstruction of two temples with the money he earned in Nepāla. In Nepāla, Rudraśiva is said to have initiated several princes into his tradition, and mention is made of the renowned Paśupatinātha temple. See, again, Bakker, Isaacson 2004, 72-4. Much more can be said, I think, about Rudraśiva and the early Nepāli kings. For an overview of the history of connections between Nepal and Kāśī, see Gaenszle 2002, 1-33. Drawing from this same inscription, Gaenszle does not mention Rudraśiva, but rather a group of unnamed 'Daśanāmi Śaṅkarācāryas', apparently taking the plural literally and not in an honorific sense, and supplementing details of their supposed affiliation (ibid., 4-5).
53 Though this matter is in dire need of systematic study, it is certain that the Siddhāntin liter-Though this matter is in dire need of systematic study, it is certain that the Siddhāntin literary output was not confined to the Siddhānta Āgamas, and it is quite likely that the Siddhāntins played a role in composing and redacting a great many Śaiva Purāṇas. The Kāśīkhaṇḍa, attributed in name to the Skandapurāṇa but clearly an independent work, certainly could have been a Siddhānta production.
In attempting to firmly make this determination we are constrained by two problems. For one, there is the problem of Purāṇic composition in general, which deliberately and systematically obscures any hints of human authorship. Secondly, the Siddhānta doctrine was a universalistic ideology, as the Siddhāntins aimed not to establish a sect, but rather to define the parameters of orthodoxy itself. As such, explicit references to identifiably Siddhāntic doctrine are not easy to find in a Purāṇic text. Despite these difficulties, it seems necessary to push the boundaries of what we can discover about Purāṇic composers, for we have too often settled for a assuming a generic 'brāhmaṇical' authorship for these texts, with analyses of sectarian affiliations rarely proceeding beyond affirming a Śaiva, Vaiṣṇava or Śākta orientation. Viśveśvara temple-providing details on not only its physical structure, but also on a ceremony that is difficult to read other than as an inauguration of a new temple and the installation of the Viśveśvara liṅga-is indication enough of this. Following Yāska, we must remain cognizant of the Purāṇa's clever manipulation of time in its account of the past. the legendary events of the 'past' are updated for the present, and at the same time these past narratives are presented as future predictions of events, with these predictions being made in divine prehistory by Śiva himself. 54 contrary to the suspicious and oversimplified characterizations of this literature as 'myth', it is evident that the Kāśīkhaṇḍa composers did not take their envisioning of the past lightly and took great care in ensuring the narrative strands crossing prehistory and history were precisely and purposefully linked to present concerns.

Conclusion
A re-evaluation of the greater Divodāsa narrative of the Kāśīkhaṇḍa is an opportunity to read the text as actively constructing its environment, rather than as an almost hopelessly flawed chronicle of history. Viewed in this way, the Divodāsa narrative has been shown to accomplish three related goals for the composers of the text. First, it submerges its own 'newness' as a text, by importing narratives such as Divodāsa from the earlier tradition of Vārāṇasī sthalapurāṇas. great importance was clearly placed on remaining true to the inherited literary tradition, even while 'renewing' it. Secondly, the Divodāsa story is updated to reflect an augmented, cosmic view of Śiva and of Vārāṇasī and also of the nature of the connection between the two. this augmented view far outstrips the comparatively tame and conventional earlier glorifications of Vārāṇasī: here, the kṣetra is not merely a favorite abode of the god Śiva, but the primordial 'place' from which he manifests the universe. This expanded understanding of the city, in the remarkably consistent logic of this carefully crafted text, leads to a corresponding expansion of the scale of the entire Divodāsa narrative itself, making it a cosmic struggle, albeit a benign one, between gods and men seeking to restore the balance of the universe. Along with this, there is the extraordinarily deft handling of Śiva's apparent absence, which becomes perhaps the primary narrative problem in reconciling the expanded Vārāṇasī as Śiva's primal abode with the basic 54 in thinking of these issues, i acknowledge that my discussion of the temporal structure of the Kāśīkhaṇḍa echoes Bakker's analysis of that of the Skandapurāṇa, even as my perspective on the text differs somewhat from his. Compare Bakker, Isaacson 2004, 6-7. Note in particular his distinguishing the introductory 'celestial Prelude', and 'Prehistory' sections of the Skandapurāṇa text from the main narratives, and also his discussion of the text's presentation of the 'legendary history' of the kṣetra, a view of the past which, while it 'does not conform to history as modern historiography would conceive it, … some agreement with the historic development of Vārāṇasī is detectable' (ibid., 6).
inherited Divodāsa narrative which has Śiva entering only after having displaced Divodāsa.
Finally, and perhaps most dramatically, the Divodāsa story is a frame for explaining the relationship between an ancient Pāśupata temple, Avimukteśvara, which had been for centuries the central Śaiva identity of the city, and a new imperial Viśveśvara temple, which in all probability was the fabled Karṇameru of the 11 th century ruler Karṇadeva. The fact that the description of the Viśveśvara temple is placed as the grand culmination of this massive text suggests that the Kāśīkhaṇḍa was composed concomitant to the construction of the temple, or shortly after its completion. And if Kalacuris were involved, there is no doubt that the production of both text and temple was being guided by the Mattamayūras, Śaiva Siddhāntin rājagurus who were rapidly promulgating the Siddhānta vision of Śaivism, not only in Vārāṇasī, but across the subcontinent. The success of the Siddhāntins, in fact, would allow them to effectively replace the Pāśupatas as the custodians of Śaiva orthopraxis in this period.
But the Siddhāntins built their doctrinal structure on Pāśupata foundations, just as their geographic spread was built upon the Śaiva networks which the Pāśupatas had established over the course of the preceding centuries. Accordingly, despite this dramatic shift in power and patronage in Vārāṇasī, the text skilfully avoids any suggestions of 'newness': the Kāśīkhaṇḍa discourse shows that the new Śaivism as not only honouring the old Pāśupata Śaivism (as is evident in its claiming that Viśveśvara worships Avimukteśvara), but also as being so rooted in it so as to be indistinguishable from it. the Kāśīkhaṇḍa handles the problems that a historical transition to a new Śaiva power centre in the city would raise for the sthalapurāṇa tradition with characteristic dexterity. Even in the supposedly supra-historical discourse of the Purāṇa, the liṅga of Avimukteśvara of Pāśupata fame is clearly marked as temporally prior to the liṅga of Viśveśvara, just as the 'historical' Avimukteśvara preceded Viśveśvara. Still, for the Kāśīkhaṇḍa, Viśveśvara, more than any other liṅga in the universe, is Śiva himself, and thus Śiva is best worshipped in Vārāṇasī in this, his most direct and powerful manifestation. the real triumph of the māhātmya, however, is that there is no suggestion that any regime change is taking place: the connection between the old and a radically new Śaivism is made nearly seamless. A Purāṇa, after all, even as it updates the past, is ultimately concerned not with transient historical events, but with eternal truths. And the truth here, according to the Kāśīkhaṇḍa, is that Śiva, lord of the universe (viśveśvara), has never abandoned his primal abode. the text thus carefully effects a subtle but unmistakable reconfiguration of the Śaiva centre, whereby the prior is honoured even as it is subsumed under a new order.
While it cannot be said that the above conclusively establishes that the Kāśīkhaṇḍa and the original Viśveśvara temple were twin projects attributable to Karṇa Kalacuri and his Mattamayūra advisers, there is as yet no convincing alternative explanation of the above-cited suggestions pointing in that direction. i believe that the preceding evaluation of the Divodāsa narrative of the Kāśīkhaṇḍa demonstrates the feasibility of this revised historical narrative.
the Kāśīkhaṇḍa, even without explicitly mentioning this Karṇameru temple as such, appears to carefully encode a legitimation of its construction by means of several intentional narrative strategies in evidence throughout the text. In this 'Purāṇa', then, the past is indeed being made new (purā navaṃ bhavati): in perfect accordance, it would seem, with the statement of Yāska well over a millennium earlier.
r e f e r e n c e s