Dholavira Held Its Rain in at Least 16 Reservoirs Cut Into Rock, and in 2021 Became India's 40th World Heritage Site

On 27 July 2021 UNESCO’s World Heritage Committee inscribed Dholavira, a Harappan city on Khadir island in the Great Rann of Kutch, as India’s 40th World Heritage Site. The city is known for its water: its builders dammed two seasonal streams and cut at least 16 reservoirs, many of them into solid rock, in a place where India’s nomination dossier puts average rainfall at 262 mm a year.

UNESCO dates the settlement to about 3000 to 1500 BCE and ranks it the sixth largest of more than 1,000 Harappan sites found so far. Most of the others, Mohenjo-daro among them (see our piece on the Indus cities and their plumbing), grew up near big rivers. The Press Information Bureau’s release on the listing says Dholavira did not, and that its place on Khadir gave access to copper, shell, agate-carnelian, steatite and lead, and to trade with Oman and Mesopotamia.

Found in 1967 to 1968, dug for 13 seasons

The annual review of the Archaeological Survey of India (ASI) for 1967 to 1968 records J. P. Joshi of its Western Circle leading a ground survey of Kutch. The team found 42 sites running from the Early Stone Age to late historical times. Eight were Harappan, two of them major settlements. Dholavira came out of that exploration. The nomination dossier India sent to UNESCO dates the find to 1968, while the ASI’s own review for 1984 to 1985 gives 1967 to 1968.

Joshi described two mounds, which he called a citadel and a lower town. R. S. Bisht of the ASI went back in 1984 to 1985 and reported more: three walled complexes, an acropolis, a middle town and a lower town, inside a rectangular fortification with gateways at regular intervals. Excavation began in 1989 to 1990, and UNESCO’s documents put Bisht’s 13 field seasons between 1989 and 2005.

Three towns inside one wall

UNESCO calls the highest and best defended part the Castle. A second walled enclosure, the Bailey, is attached to its west. North of the Castle lies a ceremonial ground that the dossier measures at 283 to 285 m east to west and 45 to 47.5 m north to south. Beyond it is the walled Middle Town, and east of that the Lower Town, which has no wall of its own but sits inside the city wall.

Between its inner corners the city wall encloses roughly 617 m by 711 m, according to the dossier’s table. ICOMOS, the body that advises UNESCO on cultural sites, reports that wall lengths and street spacing follow ratios such as 1:1, 1:2 and 5:7, and that streets were graded by width: main streets 4.2 to 5.8 m, side streets 3.2 to 3.5 m, lanes 1.2 to 1.7 m. The dossier gives the standard brick as 9 by 18 by 36 cm, a 1:2:4 ratio. Stone, not only brick, went into foundations and fortifications.

The dossier says Bisht borrowed the Rigveda’s terms parama, madhyama and avama for the three divisions, matching them to citadel, middle town and lower town. Castle and Bailey are the names the excavators gave the two halves of the citadel.

The workshops show what the city did for a living. ICOMOS lists around 12,000 beads and 1,588 drill bits among the finds, plus hundreds of seals and sealings and nearly a thousand weights made of 55 materials.

A city with 262 mm of rain and at least 16 reservoirs

The dossier is blunt about the climate. Kachchh sits on the edge of the monsoon belt, the summer rains are erratic and sometimes fail for consecutive years, there are no perennial rivers, and the groundwater is mostly brackish. For the monsoon itself, our explainer on how it works covers the machinery. Khadir, the dossier says, is the second poorest place in the region for rain, behind only Lakhpat.

The city sits between two seasonal streams, the Mansar to the north and the Manhar to the south. The dossier reports remains of dams across the Manhar at three places and across the Mansar at two. South-east of the Castle, ASI excavators exposed a check dam with an opening set high, so that surplus water, once the broad stream bed had filled, spilled into the southern chain of reservoirs. The site falls about 13 m from north-east to south-west, which the dossier calls well suited to holding water in tanks instead of letting it spread in a thin sheet that evaporates fast.

The dossier counts sixteen or more reservoirs along the north, west and south sides of the main settlement and east of the Castle. It estimates that they take up about 10 hectares, roughly a tenth of the city’s area. Two give a sense of scale:

• The east reservoir measures 73.4 m long and 29.3 m wide at the top, with walls about 7 to 7.5 m high. The rock beneath it was cut down to 10.6 m at its deepest point.
• A rock-cut reservoir south of the Castle was measured by 1999 to 2000 at 89.5 m long and 8 m deep, narrowing from 14.2 m wide at the top to 8.6 m at the base. A staircase of fifteen steps runs down its eastern side.

The engineers handled the soft limestone with care. The ASI’s reviews say they left the hard crust of the bedrock intact, because stripping it would have exposed friable stone that let water seep away, then sealed weak strata with cut stone laid in impervious clay mortar. The reviews also record that the Harappans desilted the reservoirs on a regular schedule, dumping the spoil on causeways.

Neither the dossier nor UNESCO gives a single figure for how much water the system held. The one volume number we could trace comes from Archaeology magazine, which in late 2000 reported that the reservoirs uncovered by the ASI team under Bisht together held more than 325,000 cubic yards. That is about 250,000 cubic metres on tuput’s own conversion, and the magazine’s wording makes it a minimum.

Inside the Castle, a network of covered stone drains carried monsoon runoff to a reservoir cut into the western half of the Bailey. The dossier says the larger drains were tall enough for a person to walk through and had air ducts at short intervals. The south-western corner holds a well, which the ASI review calls perhaps the largest found in a Harappan setting, about 4.25 m across and excavated to 13.6 m without reaching the water table, plus two tanks. The dossier reads the larger tank as storage and the smaller one, 4.85 m deep with a gentler staircase of 14 steps, as a possible bathing tank.

The east reservoir also contains a rock-cut well with steps. The dossier calls it a rudimentary forerunner of Gujarat’s later stepwells, Rani-ki-Vav among them, which is the State Party’s own view rather than an independent finding. It adds that Dholavira’s wells kept yielding sweet water through four years it calls the worst drought of the century, without naming the years.

The ten-sign board at the north gate

The north gate, the most elaborate in the Castle, has two raised chambers flanking a sunken passage 3.5 m wide. In the western chamber, during the 1989 to 1990 season, ASI excavators found a line of large signs made of cut pieces of a milk-white crystalline material. The review published at the time counted nine letters, each about 37 cm high and 25 to 27 cm wide. The dossier and ICOMOS later count ten, made of gypsum inlay.

The dossier gives each sign as roughly 30 by 30 cm, against the 37 cm height in the ASI’s first review. Harappa.com’s account also gives 37 cm, puts the full board at about 3 m and notes that one sign appears four times. The dossier does not reconcile the figures.

The inference about the board is the ASI’s. The line of signs matches the width of the central passage, and the marks left by a wooden frame suggest it hung above the doorway on a timber panel that rotted away. Harappa.com adds that the board fell face down, which left the gypsum pieces in place. Bisht placed it in Stage IV, the classical Harappan phase.

What it says, nobody knows. The dossier states that the meaning is unknown because the script is undeciphered. Whether the Indus signs record a language at all is itself argued between two camps. Steve Farmer, Richard Sproat and Michael Witzel published a paper in 2004 suggesting the signs were a non-linguistic symbol system, like the emblems of the ancient Near East; Farmer’s lecture slides from that year are titled The Illiterate Harappans. In 2009 Rajesh Rao, Iravatham Mahadevan and four colleagues reported in Science that the conditional entropy of Indus inscriptions, a measure of how predictable each sign is from the one before, matched linguistic systems and sat far from non-linguistic ones.

Two sets of dates, and two accounts of the end

The ASI’s sequence has seven stages. Stage I runs from about 3000 to 2900 BCE, Stage III from 2800 to 2500 BCE, Stage IV from 2500 to 2100 BCE, Stage V from 2100 to 2000 BCE, Stage VI from 1950 to 1800 BCE and Stage VII from 1500 to 1450 BCE, with the site empty from 2000 to 1950 BCE and again from 1800 to 1500 BCE. The dossier says this was built mainly on cultural remains compared with other excavated Harappan sites, then backed by radiocarbon dates on shells.

The radiocarbon work is by Torsa Sengupta and colleagues, among them Bisht and Y. S. Rawat of the ASI’s Dholavira team, in the Journal of Quaternary Science. Their abstract puts the start of occupation about 5,500 years ago, around 3500 BCE and several centuries before the ASI’s 3000 BCE, and the end of the dated sequence about 3,800 years ago, in the early Late Harappan period. The dossier cites the same paper as backing the excavator’s sequence, quoting a span from c. 3491 to 2958 BCE at the early end to before 1827 BCE at the late end. The date of the first settlers is therefore stated two ways.

On why the city declined, ICOMOS summarises the ASI’s reading. In Stage V the repairs got worse and water structures damaged mid-stage were never fixed. An earthquake near the end of that stage emptied the city for a short time. In Stage VI the settlement shrank to the Castle, Bailey and ceremonial ground. In Stage VII newcomers arrived after about 300 years with circular houses and no urban plan, and the site was never occupied again.

Sengupta and colleagues add a climate argument. They read the chemistry of fossil shells as showing that river water once reached the Great Rann in the monsoon months, and propose that a catastrophic drought at the start of the Meghalayan stage, about 4,300 to 4,100 years ago, drove the final collapse. That is the authors’ hypothesis. The ICOMOS summary of the ASI’s account mentions the earthquake and the lapse in repairs but not drought.

What UNESCO’s advisers said

India put Dholavira on its tentative list on 15 April 2014 and submitted the full dossier in January 2020. An ICOMOS technical mission visited the site from 17 to 23 December 2020, and the body approved its report on 18 March 2021.

ICOMOS backed two of the three criteria India claimed. It accepted criterion (iii), exceptional testimony to a vanished civilisation, because the site was occupied for more than 1,500 years and shows the whole arc of Harappan history. It accepted criterion (iv), an outstanding example of a type of ensemble, citing the pre-planned layout, multi-layered fortifications, reservoirs and drains, and extensive use of stone. It rejected criterion (ii), an exchange of influences, saying there was not enough evidence of how Dholavira shaped or was shaped by its trading partners. The Committee’s decision, 44COM 8B.35, inscribed the site on criteria (iii) and (iv).

The listed property covers 103 hectares. Gujarat extended the buffer zone from 148 to 4,865 hectares to take in the ancient quarries, though ICOMOS noted the extension did not yet carry legal protection for archaeological remains. The Press Information Bureau counted India’s tally at 40 properties after the vote: 32 cultural, 7 natural and one mixed, with the Ramappa temple in Telangana inscribed days earlier as the 39th. ICOMOS expected visitor numbers to rise once a new road to Bhuj cuts the drive from about six hours to about 90 minutes.