Ellora · India · 20.0268, 75.1771 · PYR record in the living atlas
Most temples are assembled. The Kailasa temple at Ellora was subtracted. Workers began at the top of a basalt hillside in Maharashtra and cut downward and inward, taking the mountain away until a complete temple was left standing in the pit they had made: courtyard, two-storey gateway, pillared halls, shrines, a separate pavilion for the bull Nandi joined to the porch by a bridge of uncut rock, life-sized elephants carved as though holding the whole thing up, and a tower rising 32.6 metres above the courtyard floor. It is one piece of the hill. The Archaeological Survey of India calls it the largest single monolithic excavation in the world, and there is no second attempt like it anywhere.
No room for error
Carving downward means every decision is final, because you cannot put stone back. The planners had to hold the finished building in their heads and in measured drawings before the first cut, and then keep several working generations aligned to that intention. The courtyard alone runs about 82 by 46 metres. Two free-standing pillars carrying flagstaffs, a Nandi pavilion built on two storeys, and the tower over the shrine all had to be left in place as the rock around them came out, in the right positions, at the right heights, first time.
The first move was to isolate the block. Three deep trenches were cut into the hillside to free a central mass of rock on three sides, and the temple was then worked out of that mass from the top down. Everything else follows from getting those trenches in the right place.
The numbers everybody repeats
Ellora is a good place to watch a figure harden into a fact. Almost every account gives the same pair: 200,000 tonnes of rock removed, in about eighteen or twenty years. Trace them and they thin out. The volume quoted in the literature is roughly three million cubic feet, about 85,000 cubic metres, and the tonnage is a conversion of that estimate rather than a measurement anyone made. Across the wider secondary literature the figures run from roughly 150,000 to 400,000 tonnes, and no published volumetric survey has ever been done to settle between them. The familiar 200,000 sits near the bottom of that spread. The duration is worse. The Kailasa temple carries no dedicatory inscription at all. There is no foundation stone, no completion date, no builder's record on the building. Every date attached to it comes from documents found somewhere else.
That problem is not confined to Kailasa. Across all thirty-four numbered caves there are fewer than about twenty datable inscriptions. That single fact is the reason the chronology of this site is soft, and it is a better reason than any general appeal to the difficulty of dating rock-cut architecture. Most of the date bands you will see assigned to individual caves are art-historical inference from style, not readings off the stone, and the patronage of particular caves is actively argued about on the same basis.
The one serious attempt to put a number on the labour is M. K. Dhavalikar's, and he was explicit that it was a model rather than a record: on his calculation, around 250 workers removing about four cubic feet each per day could have finished the temple inside roughly five and a half years, which was his argument that a single reign was enough. That is a defensible estimate honestly labelled. It is not the same thing as knowing.
Who built it, and when, is not settled
The standard attribution is to the Rashtrakuta king Krishna I, who reigned in the middle of the eighth century, and it rests on a Rashtrakuta copper-plate grant of about 812 to 813 AD which names Krishnaraja as the patron of the Kailasanatha at Elapura. Elapura is the site's own older name, and it turns up as Eloorpuram as well. The plate is the anchor for the whole attribution, and it is a document issued decades after the fact and physically unconnected to the caves.
The line usually quoted from it, in which the temple's own architect is astonished that the work was finished at all and says the gods were astonished with him, is one to handle carefully. It is repeated everywhere as though it were part of the grant. It reads much more like a later literary tradition attached to the temple than an epigraphic finding, and we would not build an argument on it. What the plate securely gives is a patron's name.
It also does not give a single reign. Inscriptions at the site support Rashtrakuta activity under at least two rulers, and subsidiary carving evidently went on after Krishna I was dead. Krishna I is best read as the king who started the project rather than the one who is the whole of it.
Two named alternatives sit on either side of that. Hermann Goetz argued in 1952 that the temple was not one campaign but a sequence, begun under Dantidurga and extended by later Rashtrakuta rulers over generations. Dhavalikar answered in 1982 that the bulk was Krishna I's, with additions afterwards. Going much further, Vedveer Arya has argued that the Kailasa predates the Rashtrakutas entirely, reading the Bagumra copper plates as describing a temple already ancient when Krishna admired it, and treating the Kailasa as the model for the Virupaksha temple at Pattadakal rather than the other way round.
Here is where the evidence currently points, and why. The architectural derivation is the strongest anchor available, and it runs from the Virupaksha at Pattadakal and the Kailasanatha at Kanchi toward Ellora, not away from it. Reversing that direction is the load-bearing move in the early-date case and it is also its weakest joint. But the honest summary is that the temple itself is silent, the inscriptions are all external, and scholars who work on the site continue to disagree about how many hands and how many reigns are in the rock.
Three religions, one cliff
Kailasa is Cave 16, and the numbering understates the place badly. Nearly a hundred caves are cut into this hill range. Thirty-four are numbered and visited: twelve Buddhist, seventeen Brahmanical and five Jain, with further groups such as the Ganesh Leni and the Jogeshwari Leni outside that sequence. The Archaeological Survey of India puts the whole span from about the sixth or seventh century AD to the eleventh or twelfth. Other summaries stretch it wider, to roughly the fifth to the twelfth centuries, and the disagreement extends to each group: the Hindu caves are given as seventh to tenth century by some and from the mid-sixth by others, and the Jain caves as ninth century onwards with an end date nobody can fix. Take all of these as ranges under discussion rather than as settled brackets.
Some of the caves are worth knowing by number. Cave 10 is a chaitya-griha called the Carpenter's Cave, whose ceiling is a ribbed barrel vault cut in imitation of timber construction, stone pretending to be joinery. Cave 32, the Indra Sabha, still holds traces of coloured plaster and has a monolithic elephant standing in its forecourt. Cave 15, the Dashavatara, carries an inscription naming the earlier Rashtrakuta ruler Dantidurga, though it does not say he commissioned the cave. Cave 15 also appears to have been altered from one tradition to another, and how common that kind of reuse was along the cliff is unresolved.
The old textbook version had tidy phases, one faith following another. Recent work has dismantled that. Shared architectural details, throne types repeated across religious boundaries and Jina images near Cave 33 dating to the seventh or eighth century point to work running simultaneously rather than in sequence, with the same craft traditions serving different patrons. The patrons were not all kings either: inscriptions in the Jain caves name donors such as Nagavarma and Sohila. Dantidurga's own copper-plate grant records him performing a ritual bath at the site around 742 AD, and a Jain inscription of 1234 AD calls the hill Charanadri and treats it as a holy place in its own right. Three religions worked the same escarpment for four centuries without erasing each other, which is at least as remarkable as the engineering.
Human hands, checked
The claims sometimes attached to Ellora, that the rock was cut with technology since lost or in an impossibly short time, come apart on the surface of the rock itself. Chisel marks cover it. The stone is Deccan Trap basalt, laid down in flows that split along predictable planes, which is precisely why this hill and not another was chosen. Unfinished and roughed-out surfaces survive across the site and show the sequence of work. Organised labour on that scale, over whatever span it actually took, is an enormous undertaking and an entirely human one. Even the Mughal emperor Aurangzeb, no friend of the temple's religion, left it standing and called it a marvellous place to walk.
What is happening to the rock now
The caves have been looked after before. In the eighteenth century the Maratha queen Ahilyabai Holkar funded restoration here, including lime-plaster work, which is one of the reasons some surfaces read the way they do.
The current problems are documented rather than speculative. The Archaeological Survey of India and UNESCO both record water seeping into the caves along the horizontal joints of the basalt, which is the same property of the stone that made it workable in the first place. Fungi and algae grow on the surfaces. Bat colonies occupy some of the caves. The surviving painted plaster, of which there is not much left, is deteriorating. A condition survey of all thirty-four caves has been carried out under the ASI's Conservation Management Plan. The engineering problem at Ellora was solved in the eighth century. The conservation problem has not been.
What our engines found
Nothing usable, and the reason is ours rather than the site's. Our alignment engine measures great-circle distance ratios between catalogued sites. Inside a 200 km radius of Ellora our atlas holds only 4 catalogued sites, and the engine could work with 3. That is far too little to test anything, so we report the gap instead of a number.
A sparse catalogue is a hole in our data. It is never evidence of absence in prehistory, and anyone who tells you otherwise is reading a limitation as a finding. The Deccan is on our roadmap, and when coverage lands this section will carry a real result you can reproduce yourself.
One limit worth stating plainly, because it is the one most often glossed over. This test compares positions between monuments. It says nothing about any building's own proportions, its height, base, slope or perimeter. That is a separate question, and at Ellora it is the interesting one: a temple carved downward from a single rock has proportions its makers had to fix in advance, which is a far better geometry question than anything we can currently test between sites.
Explore it yourself
Open the Ellora record in the living atlas to see the escarpment in 3D, place Kailasa among its neighbours across the Deccan and run the alignment and ley line engines on the real coordinates. The Ellora site-emblem tee carries the Kailasa tower, the site name and the exact coordinates in the Pyramason Chronicle style.
Sources worth reading
Goetz, "The Kailasa of Ellora and the Chronology of Rashtrakuta Art", Artibus Asiae (1952), the origin of the multi-reign argument. Dhavalikar (1982) for the answer to it and for the labour model quoted above. Walter Spink's work of 1967 onwards on the Deccan cave sequence. Owen, Carving Devotion in the Jain Caves at Ellora (Brill, 2012), which is the specialist treatment of the group that gets least attention. Burgess, Report on the Elura Cave Temples (Trübner, 1883), the first systematic survey and still worth reading for what was visible before modern conservation.