Artefacts from Ban Chiang, Thailand: Pre-metal Age cord-marked pottery

1 Introduction

Because of its extraordinary and characteristic ceramic artefacts with unique coloured patterns on Iron Age pottery and cord-marks on the surface of older Neolithic ceramics the archeological excavation site Ban Chiang in the Udon Thani province, Northeast Thailand, became a UNESCO world heritage in 1992. Intensive studies on ethnic groups living there [1], on their ceramic technology [2], on their archeological classification [3, 4] as well as on the geological setting [5], [6], [7], [8] have been published. Ban Chiang lies on the Indochina micro-plate in the northern part of the Sakon Nakhon Basin. The plate extends to the Khorat Plateau towards the south. Mineral deposits in the area result from Cretaceous granites, granodiorites and pegmatites [8]. Because of repeated flooding combined with chemical weathering of rocks the basin became an extended saline area where Tertiary marine evaporation and non-marine silica-rich sedimentation took place [5], [6], [7]. Several authors discussed possible migration and immigration in the northern-east area of today’s Thailand [1, 3, 9] using isotope analysis and radiocarbon dating of human tooth enamel and bone remains.

Artefacts of Ban Chiang are well recognizable, especially from the late Iron Age, ca. 300 BC to 200 AD. At their surface most of the Iron Age ceramics show characteristic ’zoomorphic’ design and symmetric patterns [10, 11]. The red colour consist of hematite-rich natural pigments [12] and its distinctive style may be considered as precursors of writing with some resemblance to other archeological sites [13, 14]. Ceramics from Bronze Age and early Neolithic Pre-metal Age of Ban Chiang have a different appearance. Bronze Age pottery shows a smooth brown-yellowish surface while Pre-metal Age pottery has handmade impressions, so called cord-marks at the surface also apparent e.g. in other Asian [15], American [16], Near East [17, 18] and African ceramics [19].

In previous work [20] redox-conditions and firing-temperatures had been studied using XANES amongst other techniques and it had been concluded from Fe K-edge X-ray absorption signals that Ban Chiang pottery shards were fired in all periods under reducing conditions but at different temperatures. The authors concluded that Pre-metal shards showing dark to black surface were produced at temperatures above 1300 K. Surface painting and mineral compositions had previously been analysed [12] while bulk properties like grain arrangements, mineral composition and porosity have not been investigated in detail yet. Hence, we focus in this work on compositional and morphological features of the bulk and compare Pre-metal Age pottery with a younger Bronze Age sample excavated in Ban Chiang. Apart from chemical and mineralogical techniques the volume of the samples was studied using X-ray tomographic microscopy to explore more qualitative and quantitative details of texture features of the artefacts.

2 Experimental

2.1 Samples

Both ceramic shards used in this study stemmed from Ban Chiang and had been made in different periods. The older shard PSN2-S10E13 is a Pre-metal Age cord-marked ceramic piece and has a dark grey surface (Figure 1a). The other shard 5412-S6E15 (Figure 1b) is from Bronze Age (ca. 1000–300 BC) with a yellow-brown surface which had previously been studied with focus on its surface materials [12] and redox- and firing-conditions [20]. For X-ray powder diffraction and FTIR spectroscopy ca. 3–4 mg material was removed from the bulk of the respective samples. Other techniques covered microprobe analysis and the investigation of optical thin sections prepared perpendicular to the sample surface. The chemical composition was analysed using an electron microprobe (Cameca Cambax micro-beam SEM system), technical details have been described by Beirau et al. [21]. The acceleration voltage was 15 kV, the sample current 40 nA and the beam diameter 30 μm. Table 1 shows the result of the analysis.

Figure 1:

Ceramic shards from Ban Chiang, Thailand. (a) Pre-metal Age cord-marked shard PSN2-S10E13 and (b) Bronze Age sample 5412-S6E15.

Table 1:

Electron microprobe analysis of Ban Chiang ceramics in wt.% oxides. Average chemical composition of (a) bulk of sample 5412-S6E15 (data from [12]) and (b) bulk composition of PSN2-S10E13 (this study). ^aA thin section of sample PSN2-S10E13 and line along analysed points is shown in Figure 2. Standards and standard deviations are described in [21, 22].

Sample bulk	(a) Bronze Age 5412-S6E15	(b) Pre-metal Age PSN2-S10E13
Oxide	wt.%	wt.%
Na2O	0.27	0.16
MgO	1.00	0.90
Al2O3	20.09	18.86
SiO2	59.09	63.76
K2O	1.97	1.19
CaO	0.92	0.70
TiO2	0.54	0.89
Cr2O3	<Det. lim	<Det. lim
MnO	<Det. lim	<Det. lim
Fe2O3	4.21	4.37
Total	88.15	90.87

1. ^a<Det. lim: MnO and Cr₂O₃ were below the quantitative detection limit. Total is below 100 wt.% because of pores and unknown amount of OH-groups in the samples.

Texture of the ceramics and mineralogical components of the bulk were studied using thin sections with standard thickness (25 μm), optical microscopy and using the BSE imaging technique of the electron microprobe.

The thin section (Figure 2a) shows brighter quartz grains in a dark matrix of fine clay mineral particles. A few grains show cleavage. Electron microprobe analysis of the thin section (Figure 2b) of sample PSN2-S10E13 was carried out at 100 points along a 1500 μm line starting from the sample surface towards the bulk. No chemical difference between the surface and the bulk occurred. In the bulk material of sample PSN2-S10E13, a few quartz crystals were observed like in the optical thin section (Figure 2a). The average of the microprobe measurements is shown in column b of Table 1. Chemical microprobe analysis of the younger sample 5412-S6E15 was done at 100 points along a line of 1200 μm published previously (column a of Table 1, data from [12]).

Figure 2:

Inside the Pre-metal Age shard from Ban Chiang, Thailand. (a) Thin section of sample PSN2-S10E13, unpolarized; (b) back-scattering electron image of chemically analysed points along the indicated line of the Pre-metal Age sample. Grey scale levels correspond to relative amount of back-scattered electrons where black areas (cracks, voids) indicate no back-scattering.

2.3 XRD and FTIR studies

Powder X-ray diffraction patterns were collected using a STOE powder diffractometer with Bragg Brentano geometry and Cu-Kα₁ radiation with an asymmetric Ge 111 monochromator. The diffraction patterns of material removed from the bulk of both ancient pottery fragments is shown in Figure 3. Diffraction was done in the 2-theta range 3–70° for samples PSN2-S10E13 and 5412-S6E15.

Figure 3:

X-ray powder diffraction patterns of bulk material of ceramic samples, top: PSN2-S10E13, bottom 5412-S6E15. Labelled reflections indicate minerals.

Infrared (IR) spectroscopy was performed using an N₂-purged Bruker FTIR Vertex 70 spectrometer equipped with a RT-DLaTGS detector. MIR-FIR powder infrared absorption spectra with instrumental resolution of 2 cm⁻¹ were measured using KBr pellets. Figure 4 shows IR spectra of material extracted from the shards bulk and displays the absorbance of PSN2-S10E13 and for comparison that of 5412-S6E15.

Figure 4:

IR absorbance of the bulk material of shards PSN2-S10E13 (this study) and 5412-S6E15. The upper spectrum (a) shows the entire range measured, the lower spectrum (b) shows the expanded region towards the far infrared.

2.4 X-ray tomographic microscopy (XTM)

X-ray Tomographic Microscopy (XTM) measurements were performed at beamline BL1.2W at the Synchrotron Light Research Institute (Public Organization), Nakhon Ratchasima, Thailand. The storage ring energy was 1.2 GeV, generating a ring current ranging between 150 and 80 mA. In order to get a monochromatic X-ray energy of 11.5 keV for X-ray imaging and reduced scattering as well as beam hardening artifacts, a polychromatic X-ray attenuator made of 400 μm thick aluminum foil was used. The X-ray projectors of each pottery shard were collected for 180° with 0.1° angular increment to create a dataset. X-ray images were taken by 2560 × 2160 pixels CMOS camera (PCO.EDGE 5.5, Optique Peter, Lentilly, France) with 3.61-μm resolution. The data were pre-processed and reconstructed in 3D based on a filtered-back projection algorithm using Octopus Reconstruction software (Octopus 8.9.4, TESCAN-XRE, Ghent, Belgium). Volume visualization, sculpting, and transfer function blending of the 3D XTM was processed using the open source software Drishti version 2.6.4 [26]. The value-gradient magnitude (2D) histogram was used for setting different transfer functions [26, 27], including the lowest density peak representing pore topology, middle density peak representing clay pottery fabric and highest density peak representing mineral inclusions (Figures 5 and 6c–e). Transfer functions map voxel information to optical properties and allow to distinguish between homogeneous regions and their neighbourhood by using colour information to indicate volumetric data, respectively density gradients [26]. Hence, these transfer functions were digitally coloured green (pores), dark yellow to dark brown (clay particle fraction) and red (mineral inclusions). The volume analysis of pores and mineral inclusions was performed using the Octopus image program.

Figure 5:

XTM volume analysis of shard PSN2-S10E13 using the Drishti program, using different transfer functions. (a) Original ceramic, (b) micro-tomographic surface, (c) pottery fabric (clay matrix) in the middle absorption region, (d) pores topology in the low absorption region, (e) mineral inclusions in the high absorption region. The different transfer functions were selected in specific regions of the value-gradient magnitude (2D) histogram presented in the bottom row.

Figure 6:

XTM volume analysis of pottery shard 5412-S6E15 using the Drishti program, using different transfer functions. (a) Original ceramic, (b) micro-tomographic surface, (c) pottery fabric in the middle absorption region, (d) pores topology in the low absorption region, (e) mineral inclusions in the high absorption region. The different transfer functions were selected in specific regions of the value-gradient magnitude (2D) histogram presented in the bottom row.

3 Discussion

From burial contexts, non-burial artefacts, cross-dating, samples excavation depths and their surface appearance Ban Chiang pottery had been arranged chronologically [4, 20, 28, 29]. Macroscopic and local chemical and physical probes can be used on ancient pottery to identify its mineralogical composition in order to determine which raw materials were used in certain ceramic bodies of ancient potters and other characteristic features of artefacts produced at different times. Ban Chiang artefacts from the Bronze Age (ca. 1000–300 BC) like sample 5412-S6E15 have a design free surface of yellow-brown colour. It was stated [30], [31], [32] that changes in the ceramic assemblages may indicate cultural changes in the historical course. A drastic change can be observed between Pre-metal Age and younger pottery from Ban Chiang. The surface of the Pre-metal Age shard PSN2-S10E13 with an age of 3609 ± 29 BP (BP = AD 1950) corresponding to 1659 ± 29 BC appears dark-grey to black and the outside of the ceramic shard shows cord-marks (Figure 1a). Mechanically applied cord marks have been found in ceramics from archeological sites at different continents [15], [16], [17], [18], [19]. Cord marks in Figure 1a cross under approximately 70° and were assumed to be made with cord or clay coils pressed or rolled onto the surface.

The raw material for ancient ceramic production in Ban Chiang seems to consists mainly of local clay mineral resources in all periods. Upon firing, they undergo characteristic reactions such as dehydroxylation, decomposition, or transformation. Figure 2 shows a thin section of the bulk of the Pre-metal Age sample with some quartz grains and a few feldspar micro-particles. Feldspar is expected to undergo hydrothermal alteration which converts it to clay during weathering. Despite weathering of sample PSN2-S10E13 over a period of ca. 3600 years, feldspar X-ray diffraction signals are still visible (Figure 3). The mineral might have been added to the matrix as temper to protect the pottery from cracking. FTIR indicates dominating quartz absorbance among other signals of phyllosilicates, alkali feldspar and weak absorption signals of mullite in the bulk material of PSN2-S10E13. IR absorbance spectra of bulk material of both shards (Figure 4) show several superimposed IR excitations of minerals they contain. Because the number of bands is a complex superposition of signals from minerals and impurities of very different quantities, they were neither de-convoluted nor quantitatively evaluated in this work. The band positions of strongest signals and shoulders observed in both shards were assigned qualitatively and are listed in Table 2. Strong IR absorbance peaks result from A₂ and E modes of quartz (near 370 cm⁻¹, 397 cm⁻¹, 460 cm⁻¹, 495 cm⁻¹, 694 cm⁻¹, 778 cm⁻¹, 797 cm⁻¹, 1083 cm⁻¹ and 1168 cm⁻¹) [33, 34]. Further bands which appear in both spectra come from A₁ and E₁ modes of alkali feldspars (albite, K-feldspar) as shoulders near 329 cm⁻¹, 580 cm⁻¹ and an intensive band near 1045 cm⁻¹ [35, 36] as well as signals of possibly different phyllosilicates (muscovite, biotite, montmorillonite) around 3680 cm⁻¹ (OH – stretching), 3440 cm⁻¹ (adsorbed water), 1636 cm⁻¹ (adsorbed water), 1035 cm⁻¹ (ν Si–O–Si), 622 cm⁻¹ (ν (Si–O–Si)), 526 cm⁻¹ ((δ (Si–O–Si))) and 469 cm⁻¹ ((δ (Si–O–Si))) [37, 38]. Phyllosilicate signals shown in Figure 4a are weaker in shard PSN2-S10E13 than those of 5412-S6E15. The IR spectrum of the Pre-metal Age shard PSN2-S10E13 shows weak and broad bands of mullite near 565 cm⁻¹, 740 cm⁻¹ and 870 cm⁻¹ [39] which is confirmed by XRD reflections of mullite displayed in Figure 3.

While the material of the Bronze Age shard 5412-S6E15 was found to contain some amount of calcite at the surface [12], no calcite could be detected in the bulk of both samples investigated in this work. Dominating X-ray powder diffraction intensities of quartz occur in the bulk of both shards (Figure 3). Sample PSN2-S10E13 shows additional weak signals of a minor phase of mullite near 2-theta 23.2°, 40.9° and 60.6° as well as those of alkali-feldspar.

The 3D image of pottery shards from both periods show their surface details and inner structural details on a macroscopic scale. The outer surface of PSN2-S10E13 presents an image of exterior press pattern of the pottery by hand-made rolling instruments (Figure 5b). The inner volume was fractionated by selecting the specific area density of the 2D histogram. The clay pottery (Figure 5c) appears to be heterogeneous, however PSN2-S10E13 seems to show oriented ceramic texture probably due to hand forming (Figure 5c). The pottery shard 5412-S6E15 (Figure 6c) has a less pronounced orientation of fine particles than PSN2-S10E13.

The pores topology presented by green colour in the image volume fractionated from the lowest density regions (Figures 5d and 6d) display significant differences between shards of the two periods. The PSN2-S10E13 shard contains a higher amount of pore areas than that of the 5412-S6E15 shard whereby the pore size in both shards can reach 100 μm. Table 3 shows the volume percentage of pores for each shard where the PSN2-S10E13 shard has approximately 1.23 times higher pores quantity than the 5412-S6E15 shard.

Mineral inclusions fractioned from the highest density of the 2D histogram are represented by red spots (Figures 5e and 6e). Micro-meter size crystalline particles show slightly different quantity in both shards. The PSN2-S10E13 sample contains a slightly smaller quantity of mineral particles than that of the 5412-S6E15 shard indicated by the corresponding volume percentage deduced from Octopus frame data (Table 3).

XTM studies indicate that the PSN2-S10E13 shard had a higher pore percentage than 5412-S6E15 shard. This finding supports previous work [20] that the PSN2-S10E13 shard was fired at higher temperatures than the 5412-S6E15 shard causing larger pores from the high thermal expansion and sintering of clay minerals. This is in good agreement with the voids around clay agglomerates and cavities shown in Figure 2a as well as the appearance of mullite peaks in the XRD pattern of PSN2-S10E13 (Figure 3) which are expected to appear in such a mixture of minerals at firing temperatures above 1350 K [40, 41].

The production of shards studied in this work happened in a large interval of time and in a limited geographical area. The appearance and texture of both artefacts deviates heavily and such differences had been discussed in literature e.g. see Santacreu [42] and references therein. Technological changes of pottery production might have been promoted by cultural contacts or migration, functional aspects or technological progress like a more economical firing method. It should be noted that the abandonment of the cord marks was followed by a period of pottery without surface design and more than 1000 years later in the Iron Age replaced by hematite-red painted patterns. It is still unknown if the pottery with cord mark patterns or that with coloured pictorial patterns was made to transfer information about their place of origin, however the origin of the respective pottery investigated in this work is well recognizable.

4 Conclusions

Major differences between the Bronze Age shard (5412-S6E15) and the Pre-metal Age shard (PSN2-S10E13) can be seen by the naked eye. It is the cord marked surface and dark grey colour of the ca. 3600 years old sample. Another difference elucidated using chemical and mineralogical investigations is apart from its higher content of quartz (Table 1) and its slightly lower content of phyllosilicates (Figure 4a) the presence of mullite in the Pre-metal Age shard. Optical and X-ray tomographic microscopy yield a larger porosity of sample PSN2-S10E13. The latter two findings indicate higher firing temperature (>1350 K) of sample PSN2-S10E13 compared with the younger Bronze Age shard. We do not know if the cord mark pattern at the surface of the Pre-metal Age pottery can lead to recognizability of other samples prepared in Ban Chiang during that period or in other words, it is not clear yet if the producers intended to convey group-specific aestheitc “meaning” by their patterns [43]. A comparison with cord marked pottery from other places should help to provide a better understanding of the transition of prehistoric societies from the Neolithic to the Bronze Age and beyond.