Arikamedu, on the southeast coast of India, has long been recognized as a key site in the history of the region. Just south of Pondicherry near the village of Virampattinam, its links with the Roman Empire were noted in the early twentieth century when French antiquarians formed surface collections and dug in a desultory manner (Surleau 1944: 177-97). Wheeler et al. (1946) and Casal (1949) subsequently excavated there and a new excavation was begun in 1989 under the direction of K. V. Raman and V. Begley.

It has long been known that there were beadmakers in Arikamedu, but no systematic study of their work had been conducted. The Pondicherry Museum houses a large number of artifacts from there. Most were gathered unsystematically, and the excavated material had not been properly curated. Few of the beads Whecler found could be located there or in Delhi, but a small sample given by him to the Institute of Archaeology, University College London, are still in good condition with their original labels. Several envelopes with beads excavated by Casal remain, but have deteriorated badly, some losing their identification data and others probably gone altogether.

Until the material from the new excavation is analyzed, chronological and locational data are scarce. This is regrettable, but the many beads now available and the definite patterns which they form allow us to reconstruct ancient beadmaking production. Some hints of location and chronology are to be had in what is left of Casal's material. Since 1981 the author has catalogued the beads and beadmaking refuse of Arikamedu at the Pondicherry Museum, some 49,000 pieces, 56.5 per cent of it glass, and 43.5 per cent stone. The muscum has mounted a display of them and published a visitors' guide to it (Francis 1987a). This paper dicusses this work and work with beads from other sites. It is strictly a preliminary report meant to generate hypotheses which can be tested by the new excavations at Arikamedu and other sites highlighted here. The dating used here is that tentatively worked out by Begley (1983).