Bronze Age · 1300–1046 BC · Sanxingdui, Guanghan, Sichuan, China
Silk found on 3,000-year-old bronze masks and offerings at Sanxingdui, China
Every cloth sample tested from the Sichuan sacrificial pits was silk, and the authors think it cushioned, covered or wrapped ritual offerings in the late Shang period.
Peer-reviewed paper
Sanxingdui, Guanghan, China Scraps of cloth stuck to bronzes, jades and ivories from the sacrificial pits at Sanxingdui in Sichuan, south-west China, are silk, a new study finds. Every sample the team tested proved to be silk, and some of it was dyed red. The researchers argue that the site used silk on a large scale around 3,000 years ago, most likely to cushion, cover or wrap ritual offerings.
The study was published online in Antiquity on 26 October 2026. Its authors are Jianbo Guo, Ruochen Ge, Qiutong Cai, Jingyu Tian and Lu Yang. They work at the Sichuan Provincial Institute of Cultural Relics and Archaeology, Northwest University, the Sanxingdui Museum and Yunnan University.
They argue that by the late Shang period (1300–1046 BC), silk was part of the site's sacrificial system, alongside bronze, jade and ivory.
A site with no known writing
Sanxingdui lies in western Guanghan, about 40 km from Chengdu, and covers around 12 km². It was first discovered in 1927. Its fame dates from 1986, when excavators opened two sacrificial pits, K1 and K2, and recovered more than 780 bronzes. Six more pits were excavated between 2019 and 2022, yielding gold masks and large numbers of elephant tusks.
The site's art resembles nothing made in Shang China, whose political centre lay on the Central Plains far to the north-east. Many objects were found broken or burned, which is why they are read as ritual offerings.
Later historical texts link the Shu kingdom of the Sichuan Basin to silkworm farming. There is still no direct archaeological evidence that silk was produced there at this time.
Cloth traces on 101 objects
Systematic investigation found cloth on 101 objects of bronze, gold, jade and ivory from pits K2 to K8. Most of them, 94 objects, are bronzes of 28 types. The new paper analyses a sample of these.
One large bronze mask carried three separate patches with different weaves and states of preservation. One circular ornament preserves up to five layers.
The team examined the residues under optical microscopes and a scanning electron microscope. This uses a beam of electrons instead of light to show very fine detail. The fibres were smooth, without scales or nodes, and about 10 microns (0.01 mm) across.
Their rounded triangular cross-sections match silk from the domesticated silkworm, Bombyx mori. Infrared tests showed the residue is protein. The authors ruled out animal hair, the other likely protein-based textile in ancient China, because of the tight weave and smooth fibres.
Bronze preserved most of the cloth. Corroding bronze releases copper, which slows the bacteria that would eat the fabric, and minerals gradually replace the fibres. Copper can also speed up the breakdown of silk, so its effect varies.
Other remains, such as those on an ivory fragment, are charred and brittle. The authors attribute this to prolonged heat during burial.
Ochre and a red dye
On one sample from pit K2, red specks proved to be hematite, an iron oxide that forms the pigment ochre. A yellow patch on the same piece is calcite, normally white and probably stained by soil.
Another sample has a large deep-red area where colour has soaked into the fibres. That points to an organic dye. Madder, safflower and cochineal were common red dyes in antiquity, but the authors write that identifying the one used here "requires further verification".
Fine gauze and dense silk
Most pieces are plain weave, the simplest pattern, and at least one has a twill-like pattern. Thread density varies widely. Over the eyebrow of one large mask, an open, gauze-like cloth has 12 threads per centimetre. On a circular ornament, a tight weave has 85 threads per centimetre in one direction, seven times as many.
The authors think this range shows fabrics made for different uses. Denser cloth may have served for writing, outer garments or ornaments, and lighter cloth for underwear or fine clothing. Average and maximum densities match or exceed those at two Shang sites on the Central Plains, Yinxu and Taixicun.
Laid in layers under the offerings
The excavation records show where the silk lay. In pit K3 it was always under the objects, in a band about 20 cm high in the south-central part of the floor. In K7 it lay almost always beneath them, in the north-eastern sector, about 10 cm high. The authors conclude that the silk was deliberately arranged in layers.
Spindle whorls and net sinkers found at Sanxingdui could point to local textile production. The study raises this only as a possibility.
An earlier study led by Hailing Zheng at the China National Silk Museum, published in Scientific Reports in 2024, had already detected silk at the site. It was found on a bronze grid-like object enclosing a piece of jade. The new paper covers a far larger group of objects.
Silk as a possible writing surface
The paper closes with a cautious hypothesis. Shang China had a mature writing system, but no systematic written records have been found at Sanxingdui. The authors suggest that if writing existed there, it may have been on a perishable material, and silk is a possible candidate.
Among the oldest known Chinese silk manuscripts are those from Zidanku in Changsha, dated to the turn of the 4th and 3rd centuries BC. They come from a different kingdom, Chu, almost a thousand years later. Sanxingdui has not yet yielded a single line of writing.
Sources
- Silk Found on Strange 3,000-Year-Old Bronze Masks From a Chinese Civilization With No Known Writing: Did They Write On It? (La Brújula Verde)
- >Silk and sacrifice at Sanxingdui: early evidence of... (Antiquity)
Paper: Guo J, Ge R, Cai Q, Tian J, Yang L. Silk and sacrifice at Sanxingdui: early evidence of silk use in ancient Shu, south-west China. Antiquity. Published online 2026:1-18. doi:10.15184/aqy.2026.10448
Spotted a mistake? Tell the editor about a spelling error or a fact that is wrong. We correct stories openly.
Readers' picks show on the front page.
Follow @AncientHeraldHQ on X for the day's finds as they come in.
Comments
House rules. Stay on the subject, be kind to other readers and say where your facts come from. No advertising and no personal attacks. The editor reads along and hides comments that break these rules.
Loading comments…