Sembiran and the First Indian Contacts with Bali: An Update
Ardika, I. Wayan, Bellwood, Peter, Sutaba, I. Made, Yuliati, Kade Citha, Antiquity
Further fieldwork at Sembiran, on Bali in the Indonesian archipelago, tells more about the eastern end of the exchange network running across southern Asia about 2000 years ago.
This brief note provides an account of recent excavations undertaken by Universitas Udayana and the Indonesian National Research Centre of Archaeology (Bali Branch) at the site of Sembiran in north Bali, Indonesia. The significance of this site is a very considerable one for Southeast Asian early history. It has yielded the first securely stratified evidence of Indian trade contact with Indonesia, dated to c. 2000 years ago, during the period of Rouletted Ware manufacture and Roman trade in southern and eastern India as represented by the famous site of Arikamedu in Tamil Nadu (for details and references see Ardika & Bellwood 1991; Ardika 1991).
The earlier (1987-9) excavations at Sembiran were first reported in the pages of ANTIQUITY in 1991 (Ardika & Bellwood 1991). At this time, the evidence for Indian trade contact came in the form of 79 sherds of rouletted and closely related wares of Arikamedu type (Wheeler et al. 1946: figures 12, 14), one sherd of an Arikamedu type 10 vessel, one black-slip sherd with a graffito in Brahmi or Kharoshthi script, and other black slipped sherds which might be of Indian origin (Ardika & Bellwood 1991: 2245). Neutron activation analysis showed that the Sembiran specimens of Rouletted Ware have identical pastes to samples from Arikamedu and other Indian sites (Ardika et al. 1993). All this material came from the close-set excavation squares numbered IV, VI and VII on the plan [ILLUSTRATION FOR FIGURE 1 OMITTED]. During this first excavation period a single sherd with rouletted decoration was also found in the square labelled Pacung I, about 250 m southeast of the Sembiran excavations. Indian trade wares were clearly present at Sembiran in quantity, but during this first excavation period it was unclear just how extensive the site was, or where it was located with respect to the coastline of 2000 years ago.
Since 1990 a further nine excavation squares have been dug by Indonesian archaeological teams in various locations in Sembiran (interim reports: Suastika & Yuliati 1993; Tim Peneliti 1994; Ardika et al. 1995). The squares are numbered individually in FIGURE 1 (Roman numerals IV-XIV and TPI-II). Four squares (VIII to XI) dug within 100 m of the modern beach produced no evidence of Indian pottery at all, only younger materials. This circumstance, together with the configuration of the contours on the coastal plain in the vicinity of the Sembiran excavations (Ardika & Bellwood 1991: figure 1 lower map), indicates that the archaeological site was at the head of a small sheltered embayment which no longer exists, the coastline being almost straight today. Squares XIII and XIV, both excavated to basal sterile beach sand a little below the artefact-bearing layers, indicate that the beach was over 200 m inland from its present location just prior to the ancient occupation.
Rice phytoliths have been identified by Doreen Bowdery in the Sembiran sediments; it is possible that rice agriculture was carried out inland from the site during the period of occupation. This could have led, through soil erosion and consequent alluvial deposition, to the encroachment of the coastal plain seawards by about 100 m during the occupation phase, from at least the vicinity of the inland squares XII and XIV to a well about 25 m seaward of TPI (i.e. across the extent of the area of Indian pottery distribution outlined in [ILLUSTRATION FOR FIGURE 1 OMITTED]).
The greatest density of Indian pottery covers an area about 130 m long by 100 m wide, perhaps located on the western side of a former small stream which once ran through the coastal plain in the vicinity of square XIV, which apparently shows signs of fluvial disturbance in its sediments. …