Kailasa Temple, Ellora: who carved it, how a temple is cut top-down from one rock, and the myths

Who carved the Kailasa temple at Ellora, and how? The Kailasa, Cave 16 at Ellora in Maharashtra, was carved under the patronage of the Rashtrakuta king Krishna I in the second half of the eighth century CE. The evidence is a copper-plate land grant of 812–13 CE, issued by his relative Karka II, which praises a wondrous Shiva temple at Elapura (Ellora). The Archaeological Survey of India (ASI) dates Krishna I's reign to 756–775 CE, while the archaeologist M. K. Dhavalikar gives 757–772. The method was unusual even for Ellora. Instead of digging a cave into the hillside, masons cut deep trenches to separate a huge block of basalt from the hill, then carved that block into a free-standing temple, working from the top down. Whether work began under Krishna's predecessor Dantidurga, and how long it took, are genuine scholarly debates. Claims that it took "200 years", that it could not have been done by hand, or that it needed "alien" or lost technology are not supported by the evidence.

Kailasa in one paragraph

Ellora's hill holds more than 100 caves, of which 34 are commonly visited. According to the ASI, Caves 1 to 12 are Buddhist, 13 to 29 are Brahmanical (Hindu) and 30 to 34 are Jain, and together they span roughly the sixth to the eleventh century CE. The group was inscribed as a UNESCO World Heritage site in 1983. Kailasa is the largest of them. The ASI describes an entrance gateway, a Nandi shrine, the temple itself on a massive plinth carved with elephants and lions and with Ramayana and Mahabharata panels, and cloisters around the courtyard. The main shrine rises about 23 m above the plinth. The ASI says the temple broadly follows the Dravida (South Indian) style, and Dhavalikar called it "almost an exact copy" of the Virupaksha temple at Pattadakal in Karnataka, at nearly twice the size. (For the styles themselves, see Hindu temple architecture: nagara, dravida and vesara.)

Who patronised it: the inscription and the debate

What the inscription says. The key document is the Baroda copper-plate grant of Karka II, of the Gujarat branch of the Rashtrakutas. It is dated Saka 734 (812–13 CE) and was found while digging foundations for a house in Baroda (Vadodara). It credits Krishnaraja, Krishna I, with the temple at Elapura. In Dhavalikar's translation of its verses, the gods, astonished, take the temple for the dwelling of the self-born Shiva rather than something made by hand. The architect himself wonders how he managed it, saying that the utmost perseverance would fail to do such a work again. The ASI also cites this grant and the Rashtrakuta inscriptions for Ellora's antiquity.

Why some scholars doubted a single reign. Older scholars found it hard to believe that so vast a work fitted into one king's reign of about fifteen years. James Fergusson and James Burgess called Kailasa "a miracle of patient industry". Hermann Goetz argued that it grew in eight phases, from the eighth to the thirteenth century, beginning under Dantidurga, the dynasty's founder. Doris Chatham proposed a five-phase sequence running from before Dantidurga to Krishna I's successor Dhruva. Cave 15 next door carries an inscription praising Dantidurga and recording the Rashtrakuta genealogy (ASI).

Dhavalikar's reading. In his 1983 book on the Kailas, which expands his earlier studies, Dhavalikar argued against the long-phase view. He held that the main temple, its gateway, the Nandi mandapa, the courtyard elephants and the victory pillars were planned as one design and carved under Krishna I. He also argued that Dantidurga, who spent his reign at war, had no real share in it. But he agreed that some parts of the wider complex are later. He dated the Lankeshwar cave to around the beginning of the ninth century, the shrine of the river goddesses to the first half of the ninth century (possibly under Govinda III), and the gallery at the back to the late ninth century. He also judged the famous panel of Ravana shaking Mount Kailasa to be a later addition.

So the honest summary is this: the main temple is credited by inscription to Krishna I, the start date is debated, and the complex around it kept receiving additions for roughly a century or more.

How a temple is cut from the top down

A normal temple is built upward from foundations. Kailasa had to be made in reverse. The ASI describes it as a departure from earlier convention: a huge mass of rock was first separated from the parent hill and then carved into a temple from top to bottom. In outline:

1. Choose the rock. The masons picked a spot in the basalt hill with the right shape and sloping profile.
2. Cut the trenches. Dhavalikar describes three large trenches cut into the hill to isolate a rectangular block in the middle. In his figures, two measure about 90 × 17 × 33 m and one about 50 × 10 × 33 m, though he notes that because of the hill's slope they had to be that deep over only about half their area.
3. Carve as you descend. Carving began as soon as enough surface was exposed. Following earlier observers, Dhavalikar notes that each part seems to have been finished as the work moved downward, which removed the need for scaffolding. In his reconstruction the shikhara (tower) was finished first, followed by the upper storey before the lower, the front before the interior, and, inside the shrine, the ceiling before the floor.
4. Hollow the interiors. Halls, pillars and the sanctum were cut into the block, and subsidiary shrines were cut into the surrounding rock faces.
5. Finish the surface. The ASI notes that the whole exterior was later given lime plaster with ochre paintings in the time of Ahilyabai Holkar (eighteenth century). Traces of older paintings survive in a few places, such as the portico ceiling.

Dhavalikar also quotes the art historian Benjamin Rowland's point that quarrying a temple out of a hill probably took less work than cutting, moving and assembling the stone needed to build one of the same size.

How long did it take? Estimates versus viral claims

Claim	Status	What the evidence says
Carved under Krishna I	Documented	The Baroda copper-plate grant (812–13 CE) credits Krishnaraja. The ASI and Dhavalikar both accept this.
Work began under Dantidurga	Debated	Goetz and Chatham proposed this. Dhavalikar rejected it.
Main temple done in about 10–12 years	One scholar's estimate	Dhavalikar's own, explicitly called "hypothetical": he estimated that 250 labourers cutting about 4 cubic feet each a day could dig the isolating trenches (about 1.5–2 million cubic feet of rock) in about five and a half years, with carving going on alongside.
"It took 200 years" or "many generations"	Not supported for the main temple	No source we found supports 200 years for the main temple. Goetz did spread the whole complex over several centuries, and later additions (Lankeshwar, the river-goddess shrine, the back gallery) do extend into the ninth century. The viral figure seems to blur those two things.
"Over 200,000 tonnes of rock removed"	Widely repeated, source unknown	We found the figure repeated for decades, usually as an "estimate", but not traced to a published survey. Dhavalikar gave a volume estimate (1.5–2 million cubic feet), not a tonnage.
"No modern machinery"	True, and expected	The temple is eighth-century work by stone-masons and sculptors. Doing it with hand tools is exactly what makes it remarkable.
"Impossible for humans" or "alien technology"	No evidence	The inscription credits a king and a human architect. Scholars have shown how the work could be organised, and its style ties it to the Chalukya-era Virupaksha temple. Nothing in the record calls for an outside explanation.

The legend of Kokasa and the queen's vow

A medieval tradition gives the temple a story of its own. Dhavalikar summarises it from Marathi sources, including the Verul Sivalaya Mahatmya of 1780. A king of Alajapura (modern Ellichpur) was cured of an illness after bathing at Mhaismal near Ellora. His queen had vowed to build a Shiva temple if he recovered, and to fast until she saw its shikhara. Every builder said a temple with its tower would take at least sixteen months. One master, Kokasa of Paithan, promised that she would see the tower within a week, and he did it by starting at the top of the rock. In this telling the temple was called Manikeshwar after the queen.

This is legend, not history, but it carries a real memory: that a rock-cut temple begins with its tower. Dhavalikar also notes that medieval Marathi literature remembers Kokasa as a master craftsman, and that inscriptions in Central India from the eleventh to the thirteenth century name sculptors "of the family of Kokasa". He suggested Kokasa may have been Kailasa's chief artist. That is a scholarly guess, not a documented fact. The copper-plate grant praises the architect but does not name him.

Visiting

The caves are managed by the ASI's Aurangabad Circle. Ellora village (Verul) is also home to the Grishneshwar temple, one of the twelve Jyotirlingas, which many visitors combine with the caves. Check the ASI's current timings, ticketing and weekly closing day before you go.