HistoricalChristian.Faith

Is the Kailasa temple at Ellora far older than its eighth-century date?

Type: supporting argument (off-timeline) · Assessment: weak — as the claim that the monolith predates the Rashtrakutas by millennia and reaches toward a lost or pre-Flood horizon. The concession owed first is real and is larger than at any other monument surveyed here: Kailasa carries no dedicatory inscription, its attribution rests on copper-plate grants issued some forty years after the reign they credit, and specialists have argued for a century about how many reigns the work spans. What defeats the antiquity reading is not the strength of that anchor but the kind of object it is. A monolithic rock-cut temple is negative space — the part of a cliff that was not carried away — so there is no free-standing predecessor to appropriate, no foundation to reuse, and no earlier phase that could survive underneath. A separable sub-claim, that the scale exceeds eighth-century capability, is rated in its own section and is also weak: the arithmetic of the excavation is ordinary, and the same cliff holds unfinished monoliths that show the method mid-execution · advanced by alternative-antiquity writers and temple-mystery documentarians

Related arguments: out-of-place artifacts (dates under skepticism) — see also whether this reconstruction's witnesses are securely dated · Liangzhu's jades and the technology they required · the "lathe-turned" pillars of the medieval Deccan · the vimanas and divine weapons of the Sanskrit epics · the Ta Prohm "stegosaurus" carving · Angkor's engineering and the antiquity of the Khmer capital · predynastic Egypt's hard-stone vases · the Great Sphinx's water-erosion redating · the Osireion at Abydos · the Giza radar scans and the "second Sphinx" · Sacsayhuamán's polygonal walls

The claim

Cave 16 at Ellora, in the Charanandri hills of Maharashtra, is not a building. It is a temple that was left behind when a hillside was taken away: three trenches were cut down into the basalt scarp to isolate a block, and the block was then worked from the summit downward into a two-storey Shiva temple with its shikhara, its Nandi pavilion, its gateway, its cloistered courtyard walls, its free-standing victory pillars and its life-size elephants — every one of those elements being rock that was never removed. The Archaeological Survey of India calls it "the largest single monolithic excavation in the world."¹ The volume taken out is put at roughly 3,000,000 cubic feet, about 85,000 cubic metres, on the order of 200,000 tonnes of basalt; Britannica's range is 150,000–200,000 tonnes, and popular accounts run as high as 400,000.²

The argument built on that begins from a genuine tension in the sources rather than from a feeling of awe. Krishna I, the Rashtrakuta king to whom the temple is credited, reigned about eighteen years. Hermann Goetz, reading the sculpture, counted a succession of distinct styles in the monument and concluded that work continued through several later Rashtrakuta reigns and beyond. If the carving really spans centuries and the credited reign spans eighteen years, the inference offered is that the attribution records something other than an act of construction. In its strongest popular form the claim is that a ruler encountered a structure already standing in the rock, endowed or completed or re-consecrated it, and had that recorded in terms his descendants' scribes turned into authorship — so that the temple itself belongs to a much older and now unknown culture, one that on this timeline's premise could reach back toward a pre-Flood world.

The evidential objection underneath it is stated cleanly by Christel Pilz, writing at Ancient Origins: the conventional date depends on copper plates naming a king, but inscriptions of this kind were made to glorify builders and donors and so "are not proof for the origin of the monuments they are referring to." A second strand, associated with the documentarian Praveen Mohan, presses the tool-marks and the geometrical regularity of the carving as evidence of implements more capable than iron chisels; the claim has not been taken up in the archaeological literature.³

The official date, and the method behind it

The method is epigraphic, with art-historical dating doing the fine work — and nothing radiometric anywhere in the chain. That should be said before anything else, because it is the honest description of a weaker anchor than the ones holding several other monuments discussed on this wiki.

Three facts fix it, and each carries a caveat that belongs to the argument rather than against it. First, the temple has no dedicatory inscription of its own. Second, the attribution comes from the Vadodara (Baroda) copper plates of Karkaraja II, dated Saka 734, about 812–813 — some four decades after Krishna I's death — which record a land grant in present-day Gujarat and in passing name Krishnaraja as the patron of a temple of Shiva at Elapura so extraordinary that celestial beings passing overhead judged it self-existent rather than made, and the architect himself "was wonder struck that he could build it." A Kannada grant found at Kadaba credits Krishnaraja similarly. Third — and this is the concession — those plates "are not physically connected to the caves, and do not date Krishnaraja's reign," and the land grants of Krishna's own successors do not mention the temple at all.⁴

The rest is stylistic, and it is genuinely contested among specialists. Goetz in 1952 argued from the sculpture that the project began under Krishna's predecessor Dantidurga — whose inscription stands in the adjacent Dashavatara cave (Cave 15) — that Dantidurga's part was suppressed after Krishna set his sons aside, and that Dhruva, Govinda III, Amoghavarsha and Krishna III each extended the complex, with the elephant-and-lion frieze added later still and Ahilyabai Holkar commissioning the last paintings in the eighteenth century. M. K. Dhavalikar in 1982 answered him: the main shrine, its gateway, the Nandi mandapa, the lower storey, the elephant-and-lion frieze, the court elephants and the victory pillars all belong to Krishna I's own reign, with the celebrated Ravana-beneath-Kailasa relief carved perhaps three or four decades later, and "no major part of the monolithic temple appears to have been an afterthought," the whole having been planned from the beginning.⁵ So the field's own range is: begun under Dantidurga or under Krishna I, largely finished in one reign or extended across four, with accretions into the eleventh century and repainting in the eighteenth.

What would have to hold true for a much older date

Three conditions, and each has real precedent somewhere in the world. They are conceded at full strength here, because an argument tested against a strawman has not been tested.

(a) The grant records appropriation or renovation, not excavation. Kings claiming predecessors' monuments is documented practice, not a fringe hypothesis: Peter Brand's survey of Egyptian usurpation lists a sphinx of Amenemhet III re-inscribed three separate times by three later kings, and notes that under Ramesses II the surcharging of statuary and reliefs became widespread — appropriation rather than erasure — with deep-cut body cartouches appearing as a countermeasure. And the formula "the king built" routinely covers rebuilding, enlarging or re-consecrating something already standing.⁶ A dedication is not automatically a construction date, and at Kailasa there is not even a dedication — only a third party's later report.

(b) The stylistic dating is circular or mis-anchored. Where a monument is dated by its style and the style is dated by monuments, the reasoning closes on itself. Goetz's own reading concedes stylistic heterogeneity; if the styles are read as strata rather than as a workshop's range, one could ask what the earliest stratum is anchored to.

(c) The monolith conceals an earlier phase. On a built temple this is the strongest of the three: reused footings and spolia are everywhere in the archaeological record, and a later structure sitting on earlier courses dates its own walls and not the platform beneath them.

Adversarial assessment: do the conditions hold here?

Usurpation needs an object to usurp, and negative-space carving does not supply one. This is the load-bearing objection and it is a point about the kind of thing Kailasa is, not about the authority of the people who date it. To usurp a monument you take something that exists — a statue, a stela, a standing temple — and put your name on it. At Ellora there was nothing standing. What the workers did was remove basalt; the temple is the remainder. For a predecessor to have "built" the Kailasa that Krishna then claimed, that predecessor would have had to cut the three isolating trenches, take out the whole 85,000 cubic metres, and carve the shrine, the pavilion, the gateway, the courtyard galleries, the pillars and the elephants — that is, do the entire work. The mechanism that is supposed to transfer the monument to an older culture turns out to require that the older culture perform every act the argument was trying to reassign. Usurpation is a strong tool against a free-standing built monument; against a monolithic rock-cut excavation it has nothing to hold.

There is no substrate, so condition (c) cannot even be posed. An earlier phase survives under a built temple because a builder can leave old courses in the ground. Here every surface of the monument, inside and out, is a cut face; the courtyard floor is a cut face; the cliff walls around it are cut faces. Anything that had stood in that volume before was carried out in baskets. An "earlier phase" at Kailasa would have to be located in rock that no longer exists — which is not a hidden older temple but its opposite.

The stylistic argument points forward, not backward. This is where the antiquity reading takes its evidence from the wrong end. Goetz and Dhavalikar disagree about how many reigns the work spans, but they are arguing about accretion: later Rashtrakutas adding, an eleventh-century frieze, an eighteenth-century repainting. Multi-phase construction running from the eighth century onward is not a reason to move the first phase earlier. And no style identified in the monument is stray. The temple's ancestry is dated and local: the Pallava Kailasanathar at Kanchipuram and the Chalukya Virupaksha at Pattadakal stand upstream of it, and the absorption of both traditions' craftsmen after the Rashtrakuta defeat of the Chalukyas is the ordinary explanation for finding more than one hand at Ellora.⁷ An eighth-century Kailasa has parents and cousins. A Kailasa of remote antiquity would be a fully developed Hindu temple, in a mature Dravidian idiom, with no Hindu temples on either side of it for thousands of years.

The method is visible in the same cliff, half-done. Ellora is not a single finished object; it is thirty-four excavations in every state of completion, and the unfinished ones are the strongest evidence there is about how the finished ones were made. Cave 30, the "Chhota Kailasa," is a smaller monolithic Jain temple that repeats Kailasa's process, orientation and main components — and was abandoned incomplete. Cave 31 was never finished. In Cave 29 the rock beneath Parvati in the Andhaka panel still carries nothing but point-chisel marks, the first tool in the sequence, which is how Vidya Dehejia and the stone-carver Peter Rockwell can reconstruct carvers' working order across the subcontinent at all.⁸ A monument inherited from a lost civilization does not leave its own half-carved siblings in the next bay of the same cliff.

What the real uncertainty is, stated exactly. The genuine opening at Kailasa is which Rashtrakuta reign the work began in and how many reigns it took. That is decades — at the outside, and on Goetz's most expansive reading, a couple of centuries of additions running later. The interpretation being rated needs thousands of years earlier. The gap between the uncertainty that exists and the uncertainty the argument requires is not one that any of the three conditions can close.

The manufacturing sub-claim, kept separate

"Could eighth-century people have made this?" is a different question from "is it eighth-century?", and conflating them is how this class of argument usually goes wrong. Suppose the dating were untouched; one could still hold that the labour is impossible and that the makers, whenever they worked, had tools we do not know about.

The arithmetic does not support it. Dhavalikar worked it out: at roughly four cubic feet of rock cut per person per day, 250 labourers would have finished Kailasa in about five and a half years. Multiply the crew, or divide the daily rate by three or four for hard basalt and awkward access, and the project still lands comfortably inside the eighteen-to-thirty years usually allowed — for an empire that could also field the armies and the epigraphy attesting temple programmes and sculptors' guilds across the Deccan.⁹ Two features of the technique push the same way. Subtraction skips the hardest parts of monumental building: nothing is quarried elsewhere, lifted, transported, or jointed, and there is no scaffolding problem, because working top-down means the uncut rock beneath you is the platform you stand on. And the fine carving, which is what impresses, is a thin skin over a volume that is otherwise bulk removal by hammer, chisel, wedge and lever.

The technique also has a long local pedigree at ascending scale. Rock-cut excavation in the western Deccan runs from the Buddhist viharas and chaityas of Ajanta, begun in the second century BC, through Elephanta and the early Ellora caves, to the Pallava monolithic rathas at Mamallapuram in the seventh century — free-standing temples cut from single boulders, and the direct conceptual precedent for what was done at Ellora at twenty times the size.¹⁰ Kailasa is the summit of a nine-hundred-year regional tradition, not an isolate. And the marks left by the tools are on the walls where anyone can look: the peripheral and unfinished areas at Ellora carry rough chisel work at every stage of refinement. The correct reading of the scale is that it is an extraordinary feat of organization and stamina by an imperial state — which is a real achievement, and a different thing from an out-of-place technology.

The Aurangzeb tradition, and what it is worth

The story is told everywhere: that in 1682 Aurangzeb sent a thousand men to demolish the temple, that they worked at it for three years, and that they managed to disfigure some sculpture and nothing more. It is offered as period testimony that the monument is beyond human demolition and therefore beyond human making.

It should not be used, because it does not appear to have happened. There is no historical record of Aurangzeb attempting to destroy the Kailasa temple; the tradition is a later oral one, and it circulates in mutually inconsistent versions. What the Mughal record does show is the opposite posture: Ellora was a place the emperor and other Mughal nobles went to picnic, and Aurangzeb is reported calling it a marvellous place for strolling and "one of the wonders of the work of the transcendent Artisan." The real damage at Ellora is iconoclasm of the fifteenth to seventeenth centuries, and it has the signature of that kind of attack rather than of demolition — faces, noses, breasts and limbs struck off, while the pillar-carving and the ornament beside them survive intact.¹¹ Worth adding: even if the episode were historical it would prove the hardness of Deccan basalt and the difficulty of demolition, which nobody disputes, and would say nothing whatever about when the rock was cut.

Assessment

Weak — as the claim that Kailasa is significantly older than its eighth-century date, and weak separately as the claim that its scale exceeds eighth-century capability.

The concession comes first because it is genuine and because this page would be worthless without it. Of the monuments examined in this group, Kailasa has the softest anchor: no dedicatory inscription, an attribution resting on grants issued about forty years later and physically unconnected to the site, silence about the temple in the grants of Krishna's own successors, and a century of unresolved specialist argument about which reign began the work and how many reigns carried it. Anyone who says the date is nailed down to a year is overstating. Some of the softness is even conceded from the conventional side, by Goetz, whose multi-phase reading is a scholar's result and not a sceptic's.

What the softness will not do is bear the weight placed on it, and the reason is specific to this artifact rather than borrowed from authority. The two mechanisms that genuinely relocate monument dates — a later king appropriating an existing structure, and a later structure resting on an earlier one — both need something that pre-exists the claimed builder, and a monolithic rock-cut temple is precisely the case where nothing does. The temple is the rock that was left; to attribute it to an older culture is to attribute the entire excavation to that culture, which is the conclusion rather than an argument for it, and it leaves that culture's fully mature Dravidian temple idiom standing alone in the Deccan with no antecedents and no successors for millennia while the actual antecedents — Ajanta, Elephanta, Mamallapuram, Pattadakal, Kanchipuram — sit dated and in sequence around it. Meanwhile the stylistic evidence the argument invokes runs the wrong direction, toward later additions rather than earlier origins; the labour arithmetic is unremarkable; and the method itself is on display half-finished in the adjoining bays of the same cliff. The Aurangzeb episode, the one piece of period testimony usually enlisted, is not in the record at all.

The honest report is therefore double, and the second half is the interesting one: the date of Kailasa is less securely fixed than the confident textbook sentence suggests, and the temple is nonetheless eighth-century. Its real scandal is not a lost technology. It is that a medieval Indian state, with iron and rope and baskets and perhaps a few hundred people, decided to take a mountain apart from the top down and did it.

Sources