4.1 Microstructures
Microscopic observation reveals that most of the glass beads from the Nanhai I shipwreck exhibit distinct spiral striations (Fig. 1a, j, k), particularly on the longer specimens such as NHI-3-B. The sharp little tail formed by stretching near the bead perforations pore, as seen in samples NHI-1-C, NHI-1-D, NHI-2-C, NHI-2-F, and NHI-3-E (Fig. 1a, b, e, h, l), indicates that these beads were manufactured using the coiling technique. Severe weathering is prevalent among the samples, with surfaces partially or fully covered by greyish-white weathering and corrosion products (Fig. 1a, b, c, n), and some even exhibit whitish encrustations (Fig. 1e, h), probably resulting from erosion in the marine environment. The fracture surfaces show that even beads with grey surfaces possess an opaque yellow interior, as exemplified by the sample NHI-1-D (Fig. 1b). For the black Type II samples, the interiors of the glass beads are predominantly black (Fig. 1e, h), followed by orange (Fig. 1g, j) and yellow (Fig. 1i) hues, after removing a superficial grey weathering layer.
4.2 Micro-CT analysis results and coiling forming technique
Micro-CT analysis clearly reveals the internal structural features of three representative glass beads (NHI-1-D, NHI-2-C, NHI-3-B), as shown in Fig. 2. A layer of greyish-white material is present on the surface of all three analyzed samples, appearing as a bright grey outer layer in the Micro-CT images. This greyish-white area is free of bubbles and is interpreted as a surface encrustation formed due to marine erosion.
Numerous bubbles of varying sizes are distributed throughout the three samples (Fig. 2a, c, e). Compared to Type I and Type II samples, the Type III sample NHI-3-B contains relatively fewer bubbles. The bubbles exhibit an overall preferential alignment, consistent with the direction of the glass coiling. They are predominantly oriented along the beads' axis, showing significant variation in length and diameter (approximately 0.05–0.6 mm), and displaying circular or elliptical cross-sections. Elongated, deformed bubbles, often droplet-shaped, are observed in samples NHI-2-C and NHI-3-B (Fig. 2d, f, g, h). These bubbles curve around the perforation, indicating that flow deformation occurred during the manufacturing process. Their alignment also parallels the coiling direction of the glass filament, exhibiting a characteristic layered distribution. The coiling technique involves coating a thin rod (a clay mandrel) with molten glass and continuously rotating it in one direction to form the bead. After initial shaping, while the glass remained in a molten or plastic state, the bead could be rolled on a flat surface for further thermal shaping of its form. Beads produced by this technique often retain striations perpendicular to the perforation on their surface, although these are not observable on every specimen. Subsequent thermal processing may have obliterated these striations. The use of the coiling technique by Chinese artisans for glass bead production dates back to the period from the Warring States to the Han dynasties, as evidenced by the lead-barium and potash glass beads unearthed from Han tombs in Guangzhou, which were manufactured using this method (Liu, et al. 2012).
Furthermore, bright, discrete particles are present within the glass matrix of all three samples, as shown in the magnified view in Fig. 2b. These are interpreted as devittifications within the glass, appearing in clusters, interwoven with bubbles within the glass phase (Fig. 2h). In X-ray micro-CT, brighter phases indicate higher X-ray attenuation coefficients, which typically indicate materials with higher atomic numbers or greater density. Based on this principle, it can be inferred that the crystalline phases within these beads and the material concentrated in the weathering layers probably consist of components with high atomic numbers or high density. Additionally, a layered structure can be observed in the sample NHI-2-C (Fig. 2d), where the surface layer shows higher brightness and fewer bubbles, suggesting a higher atomic number or greater density in the surface layer.
4.3 Chemical Composition
A total of 21 glass beads of the three types, including fragments NHI-1-A, B, D, NHI-2-A, and NHI-3-A, were selected for µ-XRF analysis. To understand the chemical compositional variations across different areas, analyses were typically performed on the sample surfaces, fracture sections, and areas of different colours. Furthermore, to evaluate the influence of surface weathering on flux agent losing of glass beads, the surface of the sample NHI-2-D was partially polished, allowing for a comparative analysis between the altered surface and the fresh interior material. The results are presented in Table S1. Based on the analysis of the surfaces and fracture sections of the fragment samples and NHI-2-D, it is evident that the surfaces of these beads have been severely affected by weathering. The K2O content in the grey surface areas is significantly lower than that in the internal yellow fracture sections or brown fresh surfaces. The PbO content in these grey surfaces exceeds 90%, which is attributable to surface weathering and the formation of lead-enriched secondary products caused by marine erosion. Excluding the grey surface, the main components of the beads elsewhere are PbO (28%–75%), SiO2 (11%–64%), and SnO2 (2%–10%). Additionally, the yellow areas and fracture sections of the samples contain relatively high K2O (2%–9%). The brown and yellow areas of samples NHI-2-A and NHI-2-D contain markedly higher CuO (9%–13%) than other samples, which is consistent with the layering phenomenon observed in the micro-CT analysis for bead NHI-2-C of this type, where the surface layer is lead-enriched, resulting in greater brightness in the micro-CT image compared to the interior.
Notably, relatively high SO4 contents (20%–36%) were detected in both the yellow and black surface areas of the sample NHI-2-D. For energy-dispersive X-ray fluorescence spectrometry, distinguishing S Kα (2.309 keV) from Pb Mα (2.342 keV) is challenging due to their small energy difference of only 33 eV. Given the high PbO content (34%–41%) in this sample, the potential for spectral overlap had to be considered. Inspection of the qualitative spectrum from the surface measurement points on this sample revealed that the ~ 2.342 keV peak could only be fitted by including both Pb and S, suggesting the presence of S element. In contrast, qualitative spectra from other samples where S was not detected (e.g., the surface of NHI-1-C) showed that Pb alone could fully account for the peak near 2.342 keV. After gently polishing approximately 20 µm from the surface of NHI-2-D, no sulfur was detected in the subsequent µ-XRF analysis of the exposed brown glass matrix, while the PbO and K2O contents increased significantly. This confirms the presence of SO3 on the surface of NHI-2-D, probably resulting from weathering. The black colouration of the surface layer is related to the relatively high Fe2O3 content (1.6%–5.4%). In summary, based on their composition, the beads with K₂O content exceeding 5% can be classified as potash-lead glass, primarily comprising Type II, while the other two types, with K2O content below 5%, can be classified as lead glass.
The evolution of ancient Chinese glass technology shows that from the Tang Dynasty onwards high-lead glass, due to its corrosion on the refractory crucible that is used in smelting glass, was gradually replaced by potash-lead glass. After the Southern Song Dynasty, high-lead glass almost disappeared (Gan. 2012). Because of the higher chemical stability of calcium-containing silicate glass, lead oxide was partially or fully replaced by calcium oxide from the Song and Liao periods, leading to the persistence of potash-(lime)-lead and potash-lime glasses into the Qing Dynasty (Dong et al. 2025; Zhou et al. 2022) Notably, small quantities of potash-lead glass beads (mostly opaque yellow, some Cu-containing opaque green), with 7–16% K2O and 31–38% PbO, as well as minor SnO2, appeared in the Lingnan region as early as the Han dynasties (Guangzhou Institute of Cultural Relics and Archaeology.2020). These potash-lead glass beads are believed to have been produced locally by Lingnan artisans, who adapted the techniques used to make the Indo-Pacific trade beads (such as soda-alumina and potash glass beads) that were introduced from South and Southeast Asia. This type of bead ornament probably represented a local imitation that neither gained widespread popularity nor was commonly found beyond the Lingnan region during the Han dynasties.
To investigate the weathering characteristics and element features of the glass beads further, four samples (NHI-1-A, NHI-1-B, NHI-2-A, and NHI-3-A), representing the three bead types, were selected for major and trace element analysis using LA-ICP-MS. The results (see Table 1) are consistent with the µ-XRF data obtained from sample cross-sections. The content ranges of K2O, PbO, and SiO2 are 2.71%–7.26%, 48.36%–65.64%, and 21.28%–31.16%, respectively. Type I grey-brown and Type III yellow glass beads contain relatively high levels of SnO2 (3.87%–10.65%), whereas Type II black beads exhibit high levels of CuO (~ 11%) and K2O (~ 7%), significantly exceeding that of the other two types, while their SnO2 content is low (~ 1.0%). As for trace elements, these glass beads contain elevated concentrations of As, Ag, Sb, Ni, Bi, etc. Specifically, Type I beads are richer in Pr, Nd, La, Hf, and TiO2 (Fig. 3b-3d, 3f); Type III beads have a markedly higher As content compared to the other types (Fig. 3e); and Type II blue beads show obviously higher concentrations of Ag, Sb, Ni, Bi, Zn, Co, and Au than the Type I and Type III samples (Fig. 3b, 3c, 3e, 3f). These distinctions are possibly related to the specific colorants and opacifiers used. The trace elements Co, Ni, As, Zn, Sb, Ag, and Bi in the glass beads from the Nanhai I shipwreck show broadly similar trends in variation to those observed in the bronze rings found in the same shipwreck (Chen et al. 2024) (Table S2, Fig. 4a), suggesting that both came from similar geological backgrounds or used similar sources for certain raw materials.
As indicated in Table 1, the concentrations of the lanthanide Rare Earth Element (REE) in the four glass bead samples analyzed by LA-ICP-MS are low (<1.5 ppm). The chondrite-normalized REE patterns (Sun et al. 1989) are plotted in Fig. 4b. The composition patterns of the rare earth elements in these four samples exhibit similar trends, showing right-sloping tendencies for all elements from Ce onwards. The slope of the pattern increases gradually from Ce to Eu, while the segment from Dy to Lu is relatively flat, indicating Light Rare Earth Element (LREE) enrichment relative to the Heavy Rare Earth Elements (HREE). A pronounced positive Ce anomaly ("peak") and a negative Eu anomaly ("trough") are observed. This LREE-enriched pattern is characteristic of continental crust-derived REEs, indicating a terrigenous origin for the parental rocks of the glass raw materials. Nevertheless, certain differences exist among the three types of beads, such as distinct variations in the level of Ce. Additionally, Type I has higher contents of the LREEs La, Ce, Pr, and Nd compared to the other types, while the Ce content in Type III is intermediate between the others.
Table 1
LA-ICP-MS analysis results for glass beads from the Nanhai I shipwreck
| Sample No. | Na2O | MgO | Al2O3 | SiO2 | P2O5 | K2O | CaO | TiO2 | MnO | FeO | CuO | PbO | SnO2 | Li | Be |
| wt% | wt% | wt% | wt% | wt% | wt% | wt% | wt% | wt% | wt% | wt% | wt% | wt% | ppm | ppm |
| NHI-1-A-1 | 0.05 | 0.01 | 0.07 | 26.65 | 0.00 | 3.83 | 0.01 | 0.01 | 0.01 | 0.13 | 0.05 | 65.21 | 3.96 | 4.92 | 0.21 |
| NHI-1-A-2 | 0.04 | 0.01 | 0.08 | 23.39 | 0.00 | 3.25 | 0.01 | 0.01 | 0.005 | 0.14 | 0.04 | 65.64 | 7.37 | 3.73 | 0.32 |
| NHI-1-B-1 | 0.04 | 0.01 | 0.11 | 28.70 | 0.00 | 3.90 | 0.02 | 0.01 | 0.004 | 0.11 | 0.09 | 63.14 | 3.87 | 6.21 | 0.42 |
| NHI-1-B-2 | 0.04 | 0.01 | 0.10 | 27.81 | 0.00 | 3.90 | 0.02 | 0.01 | 0.004 | 0.11 | 0.09 | 62.90 | 5.00 | 5.82 | 0.00 |
| NHI-2-A-1 | 0.12 | 0.01 | 0.12 | 29.28 | 0.01 | 6.78 | 0.04 | 0.004 | 0.01 | 0.25 | 11.19 | 51.26 | 0.93 | 7.24 | 0.29 |
| NHI-2-A-2 | 0.12 | 0.01 | 0.10 | 31.16 | 0.01 | 7.26 | 0.04 | 0.004 | 0.01 | 0.14 | 11.74 | 48.36 | 1.05 | 7.22 | 0.00 |
| NHI-3-A-1 | 0.08 | 0.01 | 0.15 | 25.57 | 0.01 | 3.21 | 0.04 | 0.004 | 0.005 | 0.11 | 0.11 | 65.03 | 5.68 | 5.19 | 0.24 |
| NHI-3-A-2 | 0.06 | 0.01 | 0.13 | 21.28 | 0.01 | 2.71 | 0.04 | 0.004 | 0.004 | 0.10 | 0.09 | 64.91 | 10.65 | 4.44 | 0.01 |
| Sample No. | B | Sc | V | Cr | Co | Ni | Zn | Ga | Ge | As | Se | Rb | Sr | Y | Zr |
| ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm |
| NHI-1-A-1 | 0.00 | 0.00 | 0.23 | 1.24 | 0.47 | 10.89 | 3.01 | 0.16 | 0.01 | 583.41 | 2.86 | 2.69 | 1.12 | 0.57 | 2.22 |
| NHI-1-A-2 | 0.62 | 0.18 | 0.67 | 0.12 | 0.85 | 14.11 | 3.03 | 0.13 | 0.09 | 450.40 | 2.82 | 2.36 | 1.09 | 0.62 | 2.22 |
| NHI-1-B-1 | 1.46 | 0.00 | 1.54 | 0.00 | 1.12 | 17.46 | 2.25 | 0.15 | 0.21 | 487.41 | 0.72 | 3.06 | 2.37 | 0.61 | 3.17 |
| NHI-1-B-2 | 0.03 | 0.15 | 1.03 | 0.31 | 1.23 | 20.95 | 1.73 | 0.17 | 0.31 | 446.30 | 1.66 | 3.05 | 2.20 | 0.60 | 3.12 |
| NHI-2-A-1 | 21.31 | 0.11 | 8.38 | 0.37 | 12.07 | 51.71 | 72.16 | 0.34 | 2.30 | 318.06 | 0.00 | 2.92 | 2.10 | 0.36 | 0.58 |
| NHI-2-A-2 | 1.56 | 0.32 | 1.38 | 0.00 | 12.49 | 55.71 | 67.00 | 0.14 | 0.61 | 316.63 | 0.36 | 3.01 | 2.53 | 0.32 | 0.58 |
| NHI-3-A-1 | 1.22 | 0.09 | 0.65 | 0.80 | 0.94 | 18.52 | 12.14 | 0.23 | 0.14 | 1085.12 | 1.01 | 3.29 | 3.05 | 0.45 | 1.06 |
| NHI-3-A-2 | 0.25 | 0.00 | 0.75 | 0.07 | 1.30 | 20.75 | 10.18 | 0.23 | 0.00 | 849.43 | 2.75 | 2.78 | 2.60 | 0.43 | 1.28 |
| Sample No. | Nb | Mo | Ag | Cd | In | Sb | Cs | Ba | La | Ce | Pr | Nd | Sm | Eu | Gd |
| ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm |
| NHI-1-A-1 | 1.19 | 0.08 | 44.46 | 0.00 | 0.00 | 30.01 | 0.02 | 4.39 | 1.04 | 2.86 | 0.22 | 0.81 | 0.13 | 0.02 | 0.20 |
| NHI-1-A-2 | 1.22 | 0.25 | 34.58 | 0.00 | 0.00 | 44.38 | 0.03 | 3.92 | 1.11 | 2.90 | 0.24 | 0.79 | 0.17 | 0.03 | 0.21 |
| NHI-1-B-1 | 1.20 | 0.16 | 38.25 | 0.00 | 0.00 | 32.86 | 0.03 | 5.00 | 1.03 | 3.00 | 0.23 | 0.85 | 0.14 | 0.02 | 0.13 |
| NHI-1-B-2 | 1.16 | 0.10 | 36.64 | 0.00 | 0.00 | 37.60 | 0.03 | 4.77 | 1.09 | 2.89 | 0.22 | 0.71 | 0.10 | 0.01 | 0.17 |
| NHI-2-A-1 | 0.45 | 0.44 | 106.19 | 0.00 | 0.67 | 208.04 | 0.05 | 5.90 | 0.38 | 1.08 | 0.08 | 0.25 | 0.07 | 0.01 | 0.07 |
| NHI-2-A-2 | 0.42 | 0.18 | 105.82 | 0.00 | 0.30 | 214.24 | 0.03 | 6.32 | 0.41 | 1.07 | 0.09 | 0.36 | 0.07 | 0.01 | 0.06 |
| NHI-3-A-1 | 0.38 | 0.31 | 56.56 | 0.00 | 0.00 | 56.12 | 0.06 | 10.00 | 0.53 | 1.76 | 0.11 | 0.43 | 0.09 | 0.03 | 0.12 |
| NHI-3-A-2 | 0.40 | 0.23 | 44.10 | 0.00 | 0.00 | 132.39 | 0.04 | 8.19 | 0.44 | 1.64 | 0.11 | 0.38 | 0.10 | 0.01 | 0.22 |
| Sample No. | Tb | Dy | Ho | Er | Tm | Yb | Lu | Hf | Ta | W | Au | Tl | Th | U | Bi |
| ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm | ppm |
| NHI-1-A-1 | 0.02 | 0.08 | 0.02 | 0.06 | 0.00 | 0.06 | 0.01 | 0.07 | 0.09 | 1.46 | 0.04 | 0.03 | 0.57 | 0.06 | 12.08 |
| NHI-1-A-2 | 0.03 | 0.08 | 0.03 | 0.08 | 0.01 | 0.06 | 0.01 | 0.09 | 0.11 | 1.39 | 0.04 | 0.03 | 0.63 | 0.06 | 19.02 |
| NHI-1-B-1 | 0.03 | 0.10 | 0.01 | 0.06 | 0.01 | 0.06 | 0.01 | 0.10 | 0.06 | 0.73 | 0.02 | 0.10 | 0.67 | 0.08 | 11.95 |
| NHI-1-B-2 | 0.03 | 0.10 | 0.02 | 0.06 | 0.01 | 0.07 | 0.01 | 0.12 | 0.08 | 0.61 | 0.05 | 0.08 | 0.65 | 0.08 | 12.03 |
| NHI-2-A-1 | 0.01 | 0.04 | 0.01 | 0.04 | 0.01 | 0.03 | 0.00 | 0.00 | 0.03 | 4.20 | 16.54 | 0.22 | 0.17 | 0.00 | 129.66 |
| NHI-2-A-2 | 0.01 | 0.05 | 0.01 | 0.01 | 0.01 | 0.03 | 0.00 | 0.01 | 0.02 | 0.55 | 5.91 | 0.24 | 0.16 | 0.05 | 124.08 |
| NHI-3-A-1 | 0.02 | 0.04 | 0.02 | 0.05 | 0.01 | 0.06 | 0.01 | 0.04 | 0.05 | 0.86 | 0.09 | 0.06 | 0.33 | 0.08 | 29.61 |
| NHI-3-A-2 | 0.04 | 0.06 | 0.01 | 0.05 | 0.00 | 0.05 | 0.01 | 0.05 | 0.05 | 0.91 | 0.11 | 0.06 | 0.28 | 0.06 | 29.21 |
Based on published data (Ma et al. 2022a; Ma et al. 2021; Xu et al. 2022; Huang 2024) we compared the results of the REE analysis of some excavated potash-containing and lead-containing glasses from China. The average chondrite-normalized REE patterns for potash glass (m-K) and lead-barium glass (Pb-Ba) from different sites show that the trend in the potash-lead and lead glass beads from the Nanhai I shipwreck is similar to that of the m-K glass, characterized by positive Ce anomalies and negative Eu anomalies. However, it differs significantly from Pb-Ba glass, which typically exhibit negative Ce anomalies and positive Eu anomalies. The variable valence elements Ce and Eu are prone to be separated from other REEs with the change of physic and chemical conditions and show abnormal features. Ce often fractionates with other trivalent elements, and under oxidation conditions, diagenetic fluids usually show negative Ce anomalies (Frimmel 2009; Kamber et al. 2001; Yang et al. 2018). For element Eu, Eu2+ is preferentially incorporated into mineral lattices at high temperatures, leading to positive Eu anomalies, whereas negative anomalies occur in low-temperature, alkaline environments (Sverjensky 1984; Yang et al. 2018). Trail et al. (2024) utilized Ce and Eu anomalies in zircon (driven by their variable valence states) to determine or investigate the oxygen fugacity (i.e., paleo-redox conditions) of the fluid during their formation. The positive Ce anomalies and negative Eu anomalies in the potash-lead glass suggest that the potassium-rich raw material was probably to have been derived mainly from felsic rocks such as granite or syenite. Such rocks typically form under relatively reduction environment and are rich in minerals like potash feldspar (e.g., microcline), mica (e.g., muscovite), or illite.
4.4 Lead Isotopes
As the abundance of the three radiogenic lead isotopes (206Pb, 207Pb, 208Pb) increased over time, deposits of different geological ages, formed from source materials derived from different reservoirs, often possessing distinct lead isotopic characteristics, which can serve as indicators of geographical origin. The lead isotopes in artifacts such as bronzes, ceramics, and glass —primarily deriving from minerals — retain the isotropic signature of the ore deposits from different primordial crustal reservoirs. Therefore, lead isotope analysis is commonly used to trace the origins of the raw material of such artifacts (Faure 2018). In this study, lead isotope ratios for glass bead samples NHI-1-A, NHI-1-B, NHI-2-A, NHI-3-A, listed in Table 2, suggest that the lead used in these samples belongs to the common lead category. In order to discuss the lead isotope characteristics of these beads, we also compare them with published data (Table S3, Fig. 5) concerning: yellow potash-lead glass beads (Tian et al. 2021), green lead-glazed ceramics of the Cizao kiln in Quanzhou, Fujian (Zhou et al. 2019), lead beads (Li et al. 2025), bronze rings (Chen et al. 2024), and bronze coins from the Nanhai I shipwreck (Ma et al. 2022b), as well as bronze coins from the Beijiao site in Nanhai (Liu et al. 2025) dating to the Northern and Southern Song dynasties. As shown in Fig. 5, the lead isotope ratios of the glass beads are strikingly congruent with those of the glazed ceramics and several bronze coins from this shipwreck and the Beijiao site—even though most of them exhibit considerable variation in lead ratios. The lead isotope characteristic of these glass beads also aligns closely with that of the “lead beads”, while differing significantly from that of the bronze rings.
Given the cargo carried by the ship, which included gold foil, silver ingots, bronze mirrors, Longquan celadon, lacquerware, and ceramic jars from the Cizao kiln in Quanzhou and Chishi kilns in Foshan, previous studies have suggested that the vessel probably utilized the monsoon winds to call at ports such as Lin'an (modern Hangzhou, Zhejiang), Mingzhou (modern Ningbo, Zhejiang), Wenzhou, the Min River Estuary, Xinghua Bay, and Quanzhou Bay, and docked for supplies and cargo at such historical ports as Fuzhou, Putian, Quanzhou, and Guangzhou (Ding 2022; Tian et al. 2024). Thus, the goods on the Nanhai I shipwreck were sourced from multiple locations. Bronze coins, which circulated as currency during the Song dynasties, originated from minting institutions in different regions (Ma et al. 2022b). However, the glass beads, lead-glazed ceramics, and lead beads could have originated from the same or similar regions and the sources of lead may have been shared among different official Song dynasty mints their production. It is noteworthy that the "lead beads" from the Nanhai I shipwreck contain small amounts of Sn and are described as "lead-tin alloy beads" (Pb > 90%, Sn < 10%). Their lead isotope ratios are similar to those of the associated potash-lead and lead glass beads, raising the question of whether these "lead beads" were used as raw materials for making the glass beads.
Table 2
Lead isotope analysis results for glass beads from the Nanhai I shipwreck
| Sample No. | 206Pb/204Pb | 2σ | 207Pb/204Pb | 2σ | 208Pb/204Pb | 2σ | 207Pb/206Pb * | 208Pb/206Pb * |
| NHI-1-A | 18.1441 | 0.0003 | 15.6393 | 0.0003 | 38.6498 | 0.0007 | 0.86195 | 2.13016 |
| NHI-2-B | 18.1439 | 0.0003 | 15.6413 | 0.0004 | 38.6547 | 0.0010 | 0.86207 | 2.13045 |
| NHI-2-A | 18.1629 | 0.0003 | 15.6457 | 0.0004 | 38.6725 | 0.0010 | 0.86141 | 2.12920 |
| NHI-3-A | 18.1437 | 0.0004 | 15.6401 | 0.0004 | 38.6517 | 0.0010 | 0.86201 | 2.13032 |
| NHI-3-A | 18.1447 | 0.0004 | 15.6406 | 0.0004 | 38.6528 | 0.0010 | 0.86199 | 2.13025 |
| NHI-3-A | 18.1447 | 0.0004 | 15.6406 | 0.0004 | 38.6528 | 0.0010 | 0.86199 | 2.13025 |
* Calculated.
(Bronze coin (N): Northern and Southern Song period coins retrieved from the Shipwreck Nanhai I; Bronze coin (B): Northern Song period coins recovered from the Beijiao shipwreck site in the South China Sea)
Lead isotope ratio analysis is currently the most widely applied method for tracing metal sources and has also been extensively used for ancient glass and ceramics. However, this method relies on the "single-source hypothesis," which posits that lead ore sources from different regions often have significantly distinct isotopic compositions. This signature does not change substantially during anthropogenic processing, thus providing an effective indicator of the link between mineral sources and artifacts. Yet, as metal and glass objects were often smelted or re-melted, the mixing of products from different sources was inevitable. Given the resulting mixed isotopic data, it is not possible to distinguish the contribution of the original copper ore from that of added lead. Even if the original lead isotopic characteristic of the copper ore were preserved, it could be overwhelmed by a high proportion of exogenous lead. This would mask the original geochemical information and prevent direct correlation with geological ore data. Without comprehensive contextual information, it is often difficult to determine whether the lead isotope ratios of an artifact reflect the geochemistry of the original ore or are merely the product of mixing processes (Budd et al. 1995). Clemenza et al. (2025) through a high-precision lead isotope and trace element analysis of the Lupa Capitolina bronze statue, clearly revealed a mixing trend during casting. They concluded that the addition of lead as a flux rendered its isotopic composition unusable for determining the origin of the bronze. Consequently, factors such as raw material mixing, recycling, remelting, and the overlapping effects on lead isotopes in ancient bronzes and glass severely limit the applicability of lead isotope analysis in provenance studies. In response, Sun et al. (2023) applied a Bayesian mixing model (MixSIAR) and Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction to lead isotope ratios in the Zhou bronzes. They discovered several mixing characteristics in bronze assemblages across periods, with primary endmembers originating mainly from five geographical units. Combined with relevant archaeological evidence, this provides clues about material sources and their evolution. For ancient Chinese native glass, strong evidence for recycling is still lacking, but lead isotopes might result from the mixing of different mineral raw materials (such as flux, colorants/opacifiers, and quartz sand). Recently, Jiang et al. (2025) building upon Sun et al.'s grouping of lead source endmembers, used a Bayesian mixing model to calculate the contribution of each endmember in lead-barium silicate products from different periods and compared the results with those from bronze artifacts of the Eastern and Western Zhou dynasties. They proposed the possibility that there could have been five relatively centralized production centers for lead-barium silicate products during the Warring States period.
Due to the abundance of coinage metal resources ("coin material") and convenient water transport, the Song government established several official mints in locations such as Yanzhou (Muzhou, modern Jiande City) in Zhejiang, Jianzhou (modern Jian'ou City) in Fujian, and Shaozhou (modern Shaoguan City) in Guangdong, such as the Shenquan Mint (神泉监), the Fengguo Mint (丰国监), the Yongtong Mint (永通监), and so on (Liu 1998; Liu 1993). If we take the Fengguo Mint in Jianzhou as an example, various bronze coins were cast during the Northern Song dynasty, such as the Tianxitongbao (天禧通宝), Xiningyuanbao (熙宁元宝), Yuanfengtongbao (元丰通宝), and Xuanhetongbao(宣和通宝) (Ma. 2024). Notably, six Northern Song bronze coins among those from the Nanhai I shipwreck and the Beijiao site, which share identical lead isotope characteristics with the glass beads, are precisely these coin types. Integrating the above information, we suggest that the lead source used for the glass beads from the Nanhai I shipwreck is similar to that used for the low-temperature lead-glazed ceramics from the Cizao kiln in Quanzhou, Fujian, and the “lead beads” found on the same shipwreck. This lead probably originated in the Nanling, South Qinling, or Middle-Lower Yangtze River regions. The beads may have been produced in coastal hubs like Fujian or Zhejiang, where metal resources were readily accessible (Li et al. 2025). As no glass workshop sites from the Southern Song period have yet been discovered, the Cizao kiln is tentatively considered a potential production kiln for these glass beads, although this warrants further investigation.
4.5 Coloring Phases and Surface Corrosion Products
The colorants and opacifiers in these glass beads were analyzed using a LabRAM Odyssey spectrometer. Raman peaks characteristic of lead-tin yellow type II (PbSn1-xSixO3 / PbSnO3) (Bagdzevičienė et al. 2011) were readily detected in the yellow areas of these glass beads, e.g., samples NHI-1-F and NHI-3-B, with characteristic peaks near 141, 267, 331, and 447 cm-1 (Fig. 6a, b). This confirms the use of lead stannate compounds as opacifiers and colorants in these yellow glass beads. Tin-based opacifiers/colorants, primarily lead-tin yellow type II and type I (PbSnO4), were first identified at the Sardis site in Turkey from the Bronze Age (8th–7th century BC) (Alicia Van Ham-Meerta et al. 2019) and became popular from 200 BC to 900 AD, emerging as a typical colorant in the Roman period. Lead-tin yellow usually produces an opaque yellow color, but when mixed with blue (colored by Cu2+), it results in an opaque green glass. Tin in ancient glass can also exist as cassiterite (SnO2), producing an opaque white; varying the ratio of cassiterite to lead-tin yellow type II alters the color and transparency of the sample. Examples of tin-based opacifiers are found in ancient Chinese glass; type II lead-tin yellow was used for opaque yellow and green, and cassiterite for opaque white, in the soda-lime glass beads dating from the Han to Song dynasties (1st century BC – 13th century AD) unearthed in Hotan, Minfeng, and Xinhe, Xinjiang (Zhao et al. 2013). Type II lead stannate was also used for opaque yellow and green colors in the soda-lime, soda-alumina, mixed-alkali, and potash-lead glass beads from the Han tombs in Guixian and Hepu, Guangxi (Zhao et al. 2013), and Guangzhou, Guangdong (Guangzhou Institute of Cultural Relics and Archaeology 2020). Furthermore, an Eastern Han opaque yellow potash glass from Nanyang, Henan, also employed type II lead stannate (Xu et al. 2022). Among these glasses using tin-based opacifiers, the soda-lime and mixed-alkali glasses were imports, while the potash-lead glass from Guangzhou was probably locally produced in Lingnan. The potash glass from Nanyang might also have originated from Lingnan. This indicates that Lingnan artisans, building upon imported lead-tin yellow opacification technology, had mastered the production of potash-lead glass using lead-tin yellow as an opacifier as early as the Han dynasty.
The surface-blackened Type II beads from the Nanhai I shipwreck mostly have brown or yellow interiors. The Raman bands at 151, 222, 406, 524, and 637 cm-1 observed in the brown interior of the sample NHI-2-D (Fig. 6c) are assigned to cuprite (Cu2O), matching the reported data (Steimecke et al. 2022; Valvo et al. 2023). This is also corroborated by the high internal CuO content (> 10%) previously determined by XRF analysis. Such cuprite colorant particles have been found in copper-red potash glass beads excavated in tombs of the Han to Six dynasties period (3rd century BC – 6th century AD) in Xinjiang (Zhao et al. 2013), Guangdong (Guangzhou Institute of Cultural Relics and Archaeology 2020), and Hubei (Zhu et al. 2012). The use of copper-based red colorants in early glass dates back to the 15th century BC (Shortland 2002). Roman craftsmen added copper to glass recipes under reducing conditions (where copper exists as Cu0 or Cu1+) to produce opaque red glass (copper-red beads) (Arletti et al. 2006). Brun et al (1991) found abundant dendritic cuprite (Cu2O) crystals and minor metallic copper (Cu0) nodules in lead-containing Celtic opaque red glass. Cuprite crystals precipitate when the glass has high CuO concentrations (5–10%) and significant lead content (lead oxide 15–50%) (Bandiera et al. 2020). The brown interior of the sample NHI-2-D, containing approximately 10% CuO and 51% PbO, suggests the possible coexistence of both metallic Cu0 and Cu2O.
Since the Song dynasty, indigenous potash-lead and potash-lime glasses popular in inland China often appeared opaque blue, opacified with fluorite (Zhou et al. 2019), thus differing significantly from the potash-lead glass beads from the Nanhai I shipwreck. This discrepancy most likely relates to the difference among consumer markets. Tian et al. (2021) suggested that the glass beads from the Nanhai I shipwreck were produced specifically for export to Southeast Asia. Judging from the flux system and opacifier/colorant tradition, they share a technological lineage with Han dynasty potash-lead glass beads from Lingnan.
Additionally, phases within the weathering layers of the glass beads were analyzed using a LabRAM XploRA Raman spectrometer. Lead white (2PbCO3·Pb(OH)2) (Burgio et al. 2001) was detected on the black surface of the sample NHI-2-H, which was identified by its characteristic Raman bands (Fig. 6d). Raman analysis was performed on weathered powder from the sample NHI-2-G (PbO > 90 wt%). The main Raman peaks of the weathering layer were observed near 1060, 973, and 443–447 cm-1 (Fig. 6e, f), with no peaks detected in the regions typical for O-H stretching or water molecule bending vibrations. The peak at 1060 cm-1 is characteristic of phosgenite (Pb2CO3Cl2), assigned to the ν1 symmetric stretching vibration of the CO32- group (Frost et al. 2003a; Frost et al. 2003b). Phosgenite can be colorless, white, yellowish-white, grey, light brown, or pale green; its presence may relate to the sample's darkish surface. The peak at 973 cm-1 is characteristic of anglesite (PbSO4), attributed to the totally symmetric Ag (ν1) vibration mode of the [SO4]2- group (Kaminskii et al. 2011). Previous studies show that Pb2CO3Cl2 is occasional in corrosion products of soil-buried lead glass, whereas PbSO4 is rarely reported. The specific weathering phase assemblage on the Nanhai I shipwreck beads is closely related to the marine environment. As the potash-lead glass beads corroded, Pb²⁺ leached out and precipitated with the abundant SO42-, CO32-, and Cl- ions in seawater, gradually forming the grey surface weathering layer. As this layer consists of lead-rich phases, it appears bright grey in Micro-CT images.