Mortimer Wheeler's 1945 Dig at Arikamedu Found Italian Red Pottery, More Than 200 Beads and a Warehouse Over 150 Feet Long at an Indian Port That Traded With Rome, and Later Excavations Dated Its First Settlement to the 3rd to 1st Centuries BCE

In April, May and June 1945 the Archaeological Survey of India dug for three months on the bank of a lagoon two miles south of Pondicherry and recovered Italian red-glazed pottery, Mediterranean wine jars, Roman lamps and a brick warehouse more than 150 feet (about 46 metres) long. The mound is known to villagers as Arikamedu, on the Coromandel coast in what is now Puducherry. Mortimer Wheeler, then the Survey’s Director General, wrote in his 1946 report that it was one of the Yavana, or Western, trading stations that Greek and Roman writers and ancient Tamil poets both describe, and the first of them identified by excavation in India.

Later excavators moved his dates back by two centuries or more. The picture they kept is of a working port that took goods in from the Mediterranean and made goods of its own, among them beads.

A red sherd in a town library

Wheeler’s own account of how the dig began is short. On a visit to Pondicherry in 1944 he saw several sherds of Italian red-glazed pottery, the so-called Arretine ware, among finds on show in the town library. Arretine is named after Arretium, the Roman name of Arezzo, and Wheeler knew it could be dated within a few decades. Here, he wrote, was a firm datum line for pre-medieval South Indian archaeology, because the Indian pottery found with it must share its date.

The mound already had a history. Wheeler cites the French traveller Le Gentil, who described the visible remains in the eighteenth century, and a record of brick-robbing as early as 1734. From 1937 G. Jouveau-Dubreuil collected beads and gems that local children picked up on the surface, one of them reported to carry the head of Augustus cut in intaglio, a carved design sunk into a stone. Brother L. Faucheux and R. Surleau, a French public works officer, dug there from 1941 without recording layers.

With the French Governor’s permission, Wheeler directed the 1945 season with A. Ghosh, Krishna Deva and a group of Survey students that included B. B. Lal and B. K. Thapar, plus about twenty-five students and staff from Indian universities and states. He called it almost the first time that modern field technique had been applied in South India.

What the trenches held

The Northern Sector held the warehouse. Wheeler described an oblong brick shell, plastered outside to keep water out, built on the old shoreline about AD 50. Its floor sat four and a half feet above the lowest course of brick, and he judged it was abandoned within a generation because the low ground flooded.

The Southern Sector stood about ten feet above flood level. Two walled courtyards there were tied to brick tanks fed and drained through culverts. Wheeler conjectured, and said it was a conjecture, that they were used to prepare muslin cloth. His report prints a letter from a government textile institute noting that masonry tanks are still used for dyeing in parts of South Arcot and Chingleput districts.

Beads came out in numbers. The report counts more than two hundred, with unfinished beads and fragments of semi-precious stone left over from the work, and Wheeler concluded they were made on a large scale at the site. The materials were carnelian, amethyst, rock crystal, agate, shell, glass and gold. Two carved gems, one of them left untrimmed, were cut by Graeco-Roman gem-cutters in his reading, which he took to suggest Western craftsmen on the spot.

Eighteen potsherds bore graffiti. Wheeler found that all but two of those that could be reconstructed were in Tamil, among the earliest known examples of the language, dated to the first and second centuries AD. Potsherds with Tamil-Brahmi writing from Keezhadi have been dated to earlier centuries, a dating that is itself argued. At Arikamedu, Wheeler recorded no coins in the ancient levels.

He drew a large conclusion from the pottery. Because the Mediterranean wares could be dated, he wrote, the Indian culture beside them “achieves the distinction of being the first ancient Indian culture to be dated archaeologically in the Indian peninsula”.

Poduke, and how firm the name is

The Periplus of the Erythraean Sea is a sailing guide by an anonymous Greek-speaking Egyptian merchant, written around the middle of the first century CE. In section 60, in Schoff’s 1912 translation, it names the most important markets on this stretch of coast as Camara, Poduca and Sopatma. Local boats worked the coast from them, larger vessels sailed for the Ganges, and “the greatest part of what is brought at any time from Egypt comes here”. Casson’s later translation spells the middle port Poduke, and Wheeler notes that Ptolemy lists a Podouke emporion north of the Kaveri mouth.

Wheeler’s appendix on the ancient name is cautious. He wrote that all that could be inferred was that Arikamedu’s position fits the indications of the Periplus and Ptolemy, 60 miles north of Tranquebar and 20 miles south of Markanam. The find of a Roman trading post beside Pondicherry, he added, gave the equation “perhaps conclusive” weight. Schoff himself had called the location of all three ports conjectural and noted that Yule preferred Pulikat. Later writers, including an online Periplus gazetteer built on Casson’s text, list Arikamedu as Poduke without that hesitation, so the identification is the working consensus rather than a proven one.

On the other side of the peninsula, the same Periplus describes the Kerala port of Muziris. The ancient Tamil poetry that Wheeler quotes, among it the epic Silappadikaram, puts Yavana merchants and sailors at Puhar and at Muziris.

Wheeler’s dates and the revisions

Wheeler dated the Arretine ware at Arikamedu to AD 20 to 50 and the first occupation to the end of the first century BCE or the start of the first century CE, leaning to the later date. His reason was partly historical: Roman trade with the East, he argued, grew out of the unification of the Mediterranean under Augustus (23 BCE to AD 14), so a Roman presence earlier than 23 BCE was unlikely. He reckoned the site was occupied for between one and two centuries, with an end around AD 200 “a reasonable supposition”.

Jean-Marie Casal dug the site again from 1947 to 1950 and identified an earlier, pre-trade settlement in the Southern Sector, which he called Megalithic. Vimala Begley then reworked both excavators’ material. Her 1983 paper in the American Journal of Archaeology argued that the site was founded around the middle of the third century BCE, that contact with the Mediterranean may have begun by the late second century BCE, and that the Arretine phase fell in the first quarter of the first century CE rather than Wheeler’s second quarter.

The Begley digs of 1989 to 1992 followed. In a chapter written in 1997 and published after her death in 2000, she set the earliest settlement, before the trade, between the third and first centuries BCE, most likely in the second. She put Mediterranean amphorae, the wine jars, before the end of the first century BCE and the Italian sigillata from about the end of that century to the middle of the first century CE, citing two other scholars’ ranges of 10 BCE to AD 30 and 10 BCE to AD 50 as approximations. The later phases ran from the second half of the first century CE into the third century or beyond, possibly into the seventh. Summaries differ on the start: ThePrint, writing in 2024, gives 300 BCE for Begley’s southern settlement, while her own chapter gives the three-century range.

Who made the red bowls

Wheeler also identified rouletted ware, a black dish whose flat inside is stamped in concentric bands of tiny dents. He believed the fine kind derived from Arretine and saw the trade as a Roman enterprise from Augustus onward. Krishna Deva, his pottery assistant, thought the coarse variety was local and was unsure where the fine variety came from.

Begley’s 1986 article in Penn Museum’s Expedition magazine looked at more than 400 rouletted sherds in the surviving Arikamedu collections. She argued that the technique was probably introduced from the classical world but that the fine ware differs from Mediterranean pottery in fabric, shape and decoration, and that skilled local potters could have made it once they understood the method. She dated its first appearance at Arikamedu to the second or first century BCE and concluded the site was a major production and distribution centre, with the ware reaching northwest Sri Lanka and the Gangetic delta. She called that a working conclusion needing further tests.

The reach went west as well. Peter Francis Jr. reports that twelve potsherds from the Red Sea port of Berenike have been identified as coming from Arikamedu or related sites, all but two of them dated to the first century BCE or first century CE. He adds that the small drawn glass beads of the Indian Ocean trade are a product of Arikamedu and associated sites, though he argues those found at Berenike were probably made at Mantai in Sri Lanka.

A site in use long after the Romans

Wheeler read Chinese celadon of the ninth to twelfth centuries and Chola coins in the disturbed upper soil as signs of medieval brick-robbing, not of settlement. Begley’s chapter reports material of the third to sixth or seventh century “slowly beginning to surface”, says medieval material is present but outside its scope, and gives no firm end date because the site is so disturbed. Francis gives the span as at least the second century BCE to the seventeenth century CE, with badly mixed layers.

What stands there now

Disha Ahluwalia, an archaeologist with the Indian Council of Historical Research, wrote in ThePrint on 7 October 2024 that the site has been under ASI protection since 1982. She reports that a 2004 conference organised by the Pondicherry administration and the Embassy of Italy proposed conservation and a UNESCO nomination under the Silk Route Sites in India project, and that the ASI acquired land and drew up a master plan with a site museum. In her account, written about twenty years after those plans were announced, a single ASI board marks the site and no signs point to it from Arikamedu Road.