Reconstruction
Sanxingdui’s fame rests on an extraordinary Bronze Age city and a group of pits filled with bronze figures and vessels, gold, jade, ivory, silk traces, and other valued materials. The walled urban core covered about 3.6 square kilometres within a roughly 12-square-kilometre archaeological site. Although commonly identified with an early Shu kingdom, Sanxingdui has yielded no deciphered contemporary texts naming its rulers or institutions. Its status as a regional political and ritual centre is therefore a strong scholarly interpretation based on settlement scale, specialized production, long-distance materials, and organized ceremonial deposition.
Food evidence is quieter than the monumental art, but it reveals the economy that made such a centre possible. Archaeobotanical studies of Sanxingdui-period contexts on the Chengdu Plain indicate an agricultural system dominated by domesticated rice, supplemented by foxtail millet and smaller quantities of broomcorn millet. Charred rice and millet remains occur at Sanxingdui itself and at related settlements. Peach stones and wild grape remains have also been reported from Songjiaheba, a settlement associated with the Sanxingdui cultural horizon. Animal teeth from Sanxingdui have been identified principally as pig and deer, suggesting pig husbandry and hunting, although the available zooarchaeological sample is too limited to quantify their dietary importance.
Exactly how these foods were prepared remains uncertain. Pottery containers and cooking or serving vessels show that grain and other foods were processed and presented, but no excavated recipe identifies a particular dish. Boiled grain, porridge, or steamed rice are reasonable reconstructions from the crops and wider Chinese cooking traditions, not direct observations at Sanxingdui. Bronze zun and related forms are conventionally interpreted as ritual drink vessels, making fermented grain beverages possible, but direct residue evidence for their contents is presently insufficient. A 2026 proteomic study identified proteins from cattle-like bovines in material from the sacrificial pits. This is direct evidence for animal offerings, not proof that the animals were eaten at a feast.
The most spectacular pits appear to represent deliberate ceremonial disposal. Objects were broken, burned elsewhere in at least some cases, arranged in layers, and buried. Current radiocarbon modelling places several pits between 1201 and 1012 BCE with high probability. Such rites converted agricultural surplus, craft labour, livestock, and imported valuables into public displays of authority and religious commitment. The exact deities, myths, and participants remain unknown.
Sanxingdui’s food significance lies less in a surviving recipe than in the relationship between intensive farming and urban power. Rice cultivation suited the wet Chengdu Plain, while millet offered ecological flexibility and reduced risk. Rice remains central to Sichuan food today, but an unbroken culinary lineage from Sanxingdui cannot be demonstrated. What endures more defensibly is the regional importance of water-managed rice agriculture and diversified cropping.
Historical context
The Shang dynasty governed major centres in the Central Plains, with Anyang serving as its late capital after about 1250 BCE. Sanxingdui developed hundreds of kilometres to the southwest as a distinct upper-Yangtze urban and ritual centre connected to wider exchange networks. Its archaeological site extends across roughly 12 square kilometres, while the reconstructed walled core covers about 3.6 square kilometres. Rice-and-foxtail-millet farming had already been established on the Chengdu Plain since the preceding Baodun period, around 2700 BCE. Near the end of Sanxingdui’s florescence, activity and many regional traditions continued at Jinsha, near present-day Chengdu.
Evidence
Archaeological
StrongExcavated city walls, buildings, workshops, pottery, and eight artifact pits directly establish a large Bronze Age settlement and major ceremonial deposits.
Archaeobotanical
ModerateCharred rice and millet from Sanxingdui cultural contexts and associated Chengdu Plain settlements support rice-dominant mixed farming, although sampling remains uneven.
Zooarchaeological
LimitedPig and deer teeth and animal representations indicate familiar domestic and hunted species, but the small studied assemblage cannot establish dietary proportions.
Residue chemical
StrongProteomic analysis has identified Bovinae-derived proteins in sacrificial-pit material, directly supporting cattle-like animal offerings; it does not establish consumption or a feast.
Written sources
LimitedNo deciphered contemporary Sanxingdui texts identify the polity, foods, rulers, or rituals. Associations with ancient Shu partly depend on later historical traditions.
Food identification
ModerateRice, foxtail millet, and broomcorn millet are supported by plant remains, while pig and deer are supported by faunal evidence. Their relative dietary importance is less certain.
Dating
StrongStratigraphy, artifact typology, refitted objects, and radiocarbon modelling securely place the principal ritual deposits in the late Shang and early Western Zhou transition, with several pits modelled between 1201 and 1012 BCE.
Preparation method
LimitedVessels demonstrate food processing and service, but specific preparations such as porridge, steamed rice, meat dishes, or fermented drinks remain reconstructions without direct recipe or residue confirmation.
Geographic attribution
StrongThe excavated site is securely located near Guanghan on the northeastern Chengdu Plain in Sichuan.
Historical interpretation
ProbableInterpreting Sanxingdui as the capital of an ancient Shu state is widely supported by its urban scale and regional centrality, but the specific political identity is not confirmed by contemporary inscriptions.
Visual reconstruction
InterpretiveReconstructions of ceremonies, clothing, food service, and the original arrangement or use of bronze figures depend on artifact context and analogy rather than direct depictions of complete events.
Sources
- 1.Yingfu Li, Tian Qiu, Jianbo Guo, and Yuniu Li (2023). New discoveries at the Sanxingdui Bronze Age site in south-west China. Antiquity 97(391), e4. doi:10.15184/aqy.2022.150Primary archaeology
- 2.Honglin Ran, Shiyu Xu, Danyang Xu, et al. (2025). Newly discovered sacrificial pits at the Sanxingdui site: Insights into Bronze Age ritual remains in Southwest China. Archaeological Research in Asia 42, 100621. doi:10.1016/j.ara.2025.100621Primary archaeology
- 3.Xuekai Qi, Ruru Chen, Chengwei Wang, et al. (2026). Protein evidence for animal sacrifice and shared ritual practices in Bronze Age Sanxingdui, Southwest China. Journal of Archaeological Science 192, 106591. doi:10.1016/j.jas.2026.106591Scientific literature
- 4.Yuanhe Yang, Haiming Li, Yong Lu, et al. (2024). The Agricultural Economy of the Sanxingdui Culture (3700–3100 BP): Archaeological and Historical Evidence from the Chengdu Plain. Land 13(6), 787. doi:10.3390/land13060787Modern synthesis
- 5.Jade d’Alpoim Guedes (2011). Millets, Rice, Social Complexity, and the Spread of Agriculture to the Chengdu Plain and Southwest China. Rice 4, 104–113. doi:10.1007/s12284-011-9071-1Scientific literature
- 6.Jade d’Alpoim Guedes, Ming Jiang, Kunyu He, et al. (2013). Site of Baodun yields earliest evidence for the spread of rice and foxtail millet agriculture to south-west China. Antiquity 87(337), 758–771. doi:10.1017/S0003598X00049449Scientific literature
Limitations
- moderatedate end
The card ends at approximately 1100 BCE, but current radiocarbon modelling gives several sacrificial pits a 95.4% probability of burial between 1201 and 1012 BCE. Because the description specifically invokes the pits, an endpoint near 1000 BCE would better encompass the evidence.
- moderatetitle
“Bronze Kingdom” and the description’s “regional state” are defensible interpretations of a large urban and ritual centre, but Sanxingdui has produced no deciphered contemporary text identifying its political structure or naming it as Shu. The kingdom attribution should not be treated as directly observed fact.
