The Turtha Canal, Cut Through Rock From the Tungabhadra Around 1400, Is Still in Use at Hampi, and Krishnadevaraya's Nagalapura Reservoir Drew 15,000 to 20,000 Workers, as Domingo Paes Recorded

A canal that takes off from a stone dam on the Tungabhadra, about a mile west of Hampi, still carries water to the fields among the ruins of the Vijayanagara capital. A. H. Longhurst of the Archaeological Survey of India described it in his Hampi handbook as running for miles, often through solid rock, along the foot of the hills. A 2019 environmental assessment prepared for the Asian Development Bank lists the same channel, spelled Turtha there, among 16 working channels of the Vijayanagara system, which serve a command area (the land the channels can water) of 11,154 hectares. The capital was founded on dry ground, and the record of how it got its water runs from a Portuguese horse dealer’s notes of the 1520s to the engineers who planned the channels’ rehabilitation in 2019.

A capital on dry ground

Vijayanagara stood in semi-arid upland Karnataka with a mean annual rainfall of about 500 mm, most of it falling between June and September, according to Kathleen Morrison and Carla Sinopoli’s 1992 study in World Archaeology. They describe rocky terrain with little arable land apart from a narrow strip of river silt along the Tungabhadra. Rice, sugarcane, tree crops and vegetables needed water all year, which in their account came mainly from river-fed canals and from tanks (reservoirs) that the canals filled. Dry crops such as millets, sorghum, oilseeds and cotton relied on rain and on wells, check-dams, terraces and small runoff-fed reservoirs.

The chronicler Fernao Nuniz, who wrote about 1535 to 1537 in Robert Sewell’s dating, says the city had once been “of no use” for want of water. He writes that it had none for gardens and orchards except from a distant source, and that the water of the country around it was brackish. His account goes on to describe how a king fixed that.

The canal cut through rock, about 1400

Nuniz says the successor of Harihara II decided to bring a large river to the city from five leagues away. He dammed it with great boulders, carried the water through the city and planted round it gardens, orchards, groves, vineyards, and plantations of lemons, oranges and roses. By Nuniz’s reckoning the water raised the king’s revenue by more than 350,000 pardaos, the gold coin in which he counts income, and, he reports, used up the treasure the king had inherited. Sewell reads Nuniz’s “Ajarao” as two rulers, Bukka II and Deva Raya I. He adds that if the channel is the one that still feeds the fields on the old city’s site, “it is a most extraordinary work”, cut for several miles out of rock at the foot of the hills.

Later writers identified it. The Madras government’s Bellary gazetteer, whose preface is dated 1904, calls it apparently the Turuttu channel, a name it translates as “swift”, and gives its length as 12 miles (about 19 km). Longhurst calls it a most remarkable work and says it was perhaps made by Bukka II. On the date, the sources vary in confidence. The gazetteer says the anicut (a low river dam) seems to be Bukka II’s, who ruled from 1399 to 1406. Two Karnataka irrigation engineers, M. G. Shivakumar and R. Cheluvaraju, give 1399 outright and note that the dam joins rocky islands already in the riverbed.

Ten dams laid across the river

The Turtha dam is one of a series. The Bellary gazetteer counts ten anicuts across the Tungabhadra in its district, each feeding a channel, with three more in ruins. It says they are made of large boulders and rough stone piled without mortar or cement, and that most run diagonally or in a zigzag across the river to use every ledge, island and boulder. The Asian Development Bank assessment, written 115 years later, counts the system differently: 17 anicuts and originally 19 channels, 18 of them in Karnataka and one supplying Andhra Pradesh.

Shivakumar and Cheluvaraju point to the planning. One channel crosses others on stone aqueducts so that it does not lose height and therefore command area. The Raya channel, which the gazetteer calls the most important of the whole series and which at 20 miles is the longest in its table, runs along the side of low hills at a height that keeps it above the bed of the tank at its end. The gazetteer quotes Major Henderson’s 1862 report: “Great science and ingenuity are displayed in the selection of the line.” Parts are cut in solid rock and parts are embanked in stone, plastered with lime and backed with earth.

Dates for the others are thinner. The Vallabhapuram dam, which feeds the Basavanna channel, was “undoubtedly” built in Krishnadevaraya’s time, the gazetteer says, because an inscription on the nearby bank is dated 1521. Shivakumar and Cheluvaraju also give 1521 for a dam about 30 km west of Hampi. The gazetteer records a local tradition that an officer named Mudda built the whole series under Krishnadevaraya, and adds that nothing is definitely known about the others.

The Kamalapuram tank and the royal enclosure

The Raya channel ends in the Kamalapuram tank, which Longhurst found full of fish, with the road to Kamalapur running along its embankment. In his day it irrigated some 450 acres of rice and sugarcane and topped up the Turuttu channel’s supply. Shivakumar and Cheluvaraju add that the village of Kamalapura was part of the capital and that the tank’s water served its houses, carried by gravity from a point high on the river. Archaeology magazine, reporting on Morrison’s fieldwork in 2009, calls it a 14th-century reservoir; the other sources used here give no founding date for the tank.

Inside the royal enclosure the water was carried in stone. Abdur Razzak, who was at the capital in 1443, noted in the palace area “numerous running streams and canals formed of chiselled stone”. Longhurst traced stone aqueducts, channels and earthenware pipes that fed small tanks and baths in the citadel, and he reasoned that the river channel lay too low to supply it, so the water was probably lifted from wells. UNESCO’s description of Hampi lists water structures among the site’s more than 1,600 surviving remains and says the temple townships had tanks built with what it calls unique hydraulic technologies. At the Vitthala temple it names a large stepped tank with a pavilion at its centre, wells and a network of water channels. The tuput guide to Hampi as a World Heritage Site covers those monuments.

Krishnadevaraya’s reservoir at Nagalapura

Domingo Paes, whose account Sewell dates to about 1522, describes a tank the king made at Nagalapura, the modern Hospet, which he had built as a new town. The tank sat at the mouth of two hills so that water running off both sides collected there, and more arrived through pipes from a lake over three leagues away. Three large carved pillars on the tank joined pipes used to irrigate gardens and rice fields, and the king had a hill broken down to make room for it. “I saw so many people at work,” Paes wrote, that there must have been “fifteen or twenty thousand men, looking like ants.” The king divided the tank among his captains, each answerable for finishing the share his people dug. Paes gave its width as a falcon shot, a measure Sewell explains as taken from an old piece of artillery.

Nuniz adds the engineering. He says the king asked the Portuguese governor at Goa for masons and got a stoneworker named Joao della Ponte. The result was a bank across the middle of a valley, a crossbow shot long and wide, with openings and pipes below through which the water was let out and closed off at will. Water from it fed channels to rice fields and gardens. Nuniz says the king gave the people the irrigated lands “free for nine years” so that they would improve them, and that the revenue from them had already reached 20,000 pardaos.

Sewell placed the lake at the mouth of a valley in the hills south-west of Hospet, where the high road to the southern districts now runs along its bank, and Longhurst quotes that identification without dispute. Both Paes and Nuniz also report that men were put to death at the site as a sacrifice to make the dam hold. That report rests on the two travellers alone, and the fuller piece on Krishnadevaraya sets it out.

Who paid for the water

Morrison and Sinopoli explain how such works were financed. Canals and several of the larger reservoirs in the capital’s region were paid for by the Vijayanagara rulers through temple investment. The person or institution that built an irrigation work was entitled to a share of the increased harvest, they write, usually a quarter, and the share went to the temple. The labour needed was large enough to produce specialist engineers and, they note citing earlier scholars, a caste of reservoir diggers. Shivakumar and Cheluvaraju write that building irrigation works was counted as religious merit second only to building temples.

The programme grew in the early sixteenth century. Morrison and Sinopoli say that royal extension of the canal system then made several small reservoirs redundant. Archaeology magazine reports that Morrison’s survey of the Daroji valley, about 10 miles south of the urban core, recorded hundreds of dams and reservoirs, with an explosion of reservoir building in the early 1500s as the capital’s population outgrew its canal-fed fields. Nearly half the land there was given to reservoirs or the fields they watered. For comparison, the Veeranam tank in Tamil Nadu shows the older Chola practice of a single great embanked reservoir.

What worked, and what the records add

The Nagalapura dam had a mixed record. Paes says it burst two or three times during construction. Nuniz, writing in the 1530s, says it was in use and paying revenue. Longhurst wrote that the work had been useless for many years. Archaeology magazine names the Rayakere dam in the Daroji valley as likely the one Paes saw: 91 feet high, 130 feet thick, more than half a mile long and faced with red stone. The magazine’s account says it never held water, probably because the ground was too porous.

The river dams fared better. The Bellary gazetteer records that they leak less than expected because silt has sealed the lower parts, and that in 1904 they irrigated some 13,200 acres (about 5,340 hectares) in Hospet taluk, 2,300 in Bellary and 900 in Adoni. The 2019 assessment says the boulder dams still leak and the channels are the subject of a rehabilitation project. Two of the old dams, Vallabhapur and Hosakote, now lie under the backwater of the Tungabhadra Dam, so the Basavanna and Raya channels are fed through a sluice in that dam. The same document finds at least 12 protected monuments along the canals in the Hampi World Heritage Area, nine of them on the Turtha canal, and records that 19.3 km of canal lie in its core zone.