1The Kushan Empire had its origins in the 2nd century BC, when a confederation of nomadic Indo-European Yuezhi tribes invaded and settled in the Sogdiana and the Bactria areas in Central Asia (Puri, 1999; Rapin, 2007). This branch of the Yuezhi had their origins in north-western China, where they lived until being pushed back by the Xiongnu. The Xiongnu, a pastoral nomadic tribal confederation from the Ordos (Thorley, 1971) who produced a large quantity of gold work (Bunker, 1993), occupied a large territory, from the Yellow River to the North of Siberia.
Figure 1: Map of Tajikistan showing the regions of Beshkent and Dangara.
Figure 1 : Carte du Tadjikistan avec la localisation des régions de Beshkent et de Dangara.
2 Only Sima Qian (145-86 BC) provides in his Shiji (Record of the Historian) some details about this civilisation and about their military superiority, which allowed them to defeat the Han and become the first powerful empire of the steppes along the Chinese frontier (Sarkhosh Curtis and Stewart, 2007).
3The branch of the Yuezhi pushed back by the Xiongnu settled along the Oxus River and in southern Bactria (present day Afghanistan and Tajikistan), but expanded from the 1st to the 4th century AD, especially under Kanishka I (early 2nd century AD), to north-western India, Pakistan, Uzbekistan and Afghanistan (Dani et al., 1996). Between the 1st and the 3rd century AD, the Kushan controlled Gandhara, where the main road connecting India with the Silk Road and the Mediterranean Sea ran (Wood, 2002; Thorley, 1971).
Figure 2: The five gold Kushan itemsselected for analysis from the collections of the National Museum of Antiquities of Tajikistan:
Figure 2 : Les cinq objets Kouchanes en or appartenant à la collection du musée National des Antiquités du Tadjikistan, sélectionnés pour analyse
(a) pendant 274 (length 24 mm, width 10 mm, weight 1.870 g); (b) pendant 267 (length 30 mm, width 19 mm, weight 3.780 g); (c) pendant earring 270 (length 55.9 mm, disc length 18.0 mm, width 10.5 mm, weight 2.483 g); (d) earring 271 (length 38.0 mm, width 10.0 mm, weight 3.447 g); (e) earring 274 (length 23 mm, width 14 mm, weight 1.654 g).
(a) pendentif 272 (longueur 24 mm, largeur 10mm, poids 1,870 g) ; (b) pendentif 267 (longueur 30mm, largeur 19 mm, poids 3,780 g ; (c) pendentif earring 270 (longueur 55,9 mm, largeur du disque 18.0 mm, width 10,5 mm, poids 2,483 g) ; (d) boucle d’oreille 271 (longueur 38,0 mm, largeur 10,0 mm, poids 3,447 g) ; (e) boucle d’oreille 274 (longueur 23 mm, largeur 14 mm, poids 1,654 g).
4The Kushan Empire was marked by a time of intensive development of the economy (Mukhamedjanov,1994) and culture, with two distinct periods separated by a ‘dark age’ (Sarianidi, 1980): the downfall of the Greco-Bactrian Kingdom during the middle of the 2nd century BC, and the powerful empire of the Great Kushan in the 1st-2nd century AD (Dani et al., 1996). Chinese, Iranian and Indian traditions were brought together under the Kushan Empire, but the Greek traditions brought by the conquest of Alexander the Great are also evident in the typology of their art. The objects from Tillya Tepe, including the gold Achaemenid type mountain goat, the aureus of Tiberius, the Mithradates II coin, the intaglios with profiles of Athena and inscriptions in Greek, the Bactrian Aphrodite, the jades from China, the garnets from India and other artefacts illustrate all these influences (Sarianidi, 1980).
5Little is known about the different goldsmithing techniques and choices from the period of the Great Kushan Empire (Drevnosti Tadzhikistana, 1985). The aim of this work is to undertake a first approach to this question by studying a few items with known provenances, originating from secure archaeological contexts situated in the most important archaeological regions in Tajikistan (Fig. 1): the Beshkent Valley (Medvedskaya et al., 1975; Sedov, 1979) and the Dangara Valley (Mandelshtam, 1968). Five items found in established excavations, representative of the Kushan jewellery and containing gold wires and granulation, were selected from the collections of the National Museum of Antiquities of Tajikistan (Vitobello et al., 2009). Three of them were found in the Beshkent Valley and are dated to a period between the 1st and the 2nd century AD: an amphora-shaped pendant and a globular openwork pendant, found in the mid 1970s during the excavation of the Beshkent -cemetery V (Figs. 2a and 2b); and a cloisonné pendant found in 1978 during the excavation of the Beshkent cemetery VII (Fig. 2c). The pendant earring (Fig. 2d) was found in the late 1970s during the excavation of the Ksirov cemetery III, mortuary enclosure 19, in the Dangara Valley and is dated to a period between the 1st century BC and the 2nd century AD. Finally, an amphora-shaped earring dated to a period between the 1st and the 2nd century AD was found in 1973-1974 during the excavation of the “Ittifoq” cemetery in the Parkhar Valley (Fig. 2e).
6Within the context of Work Package 5 of the European project AUTHENTICO, the jewellery was transported to the C2RMF in Paris for non-destructive examination by optical microscopy, SEM, and X-radiography (for details on the complementary use of these techniques see Guerra, 2008). The composition of the jewellery was determined by PIXE (particle induced X-ray emission) at the AGLAE accelerator of the C2RMF, with a 3 MeV external proton beam of 50 µm diameter and an intensity of 30-40 nA, allowing a depth of analysis which in general is sufficient to avoid the copper depletion. Simultaneous PIGE (particle induced gamma-ray emission) was carried out in order to compare the compositions of the deep and surface layers.
7The amphora-shaped earring reference 274 in Catalogue 1983 and reference 8 of Masov et al. (2005), page 120 (Fig. 2a) is composed of gold strips of about 950 µm diameter, bent, folded and joined to produce symmetric handles holding the amphora’s body. Two discs of 130-200 µm thickness separate a bead of certainly iron pyrite (this type of decoration is common for Tillya Tepe jewellery, Collectif, 2006) and pearl elements that constitute the body. The top of a rod going through the body is threaded around the amphora’s suspension hook and is finished with a pyramid of four large granules of 1650 µm diameter.
8Globular pendant reference 272 in Catalogue 1983, reference 6 of Masov et al. (2005), page 120 (Fig. 2b) consists of an iron pyrite bead covered with an open net birdcage of 500 µm diameter strip-twisted wires bent to form loops at the bottoms (Fig. 3). The junction points of the wires are ornate with small plain discs of 800 µm diameter. A suspension solid gold ring is joined to a strip of gold folded into a narrow cylindrical shape surrounded by a ringlet of granules of 650 µm diameter. These elements are joined to a disc. No solder is visible and a scan carried out by PIXE did not detect the presence of an alloy with different contents of Cu or/and Ag; however, the morphology of the joins corresponds to the use of hard soldering. A rod with seams running longitudinally along the wire (morphologically similar to the wires from pendant MK01.40.109 of Tillya Tepe, as discussed by Ogden, 2007) goes through the bead, is coiled around the lower half of the loop, and is completed with a pyramid of four granules of 1450 µm diameter. Additional smaller pyramids formed with granules of 650 µm diameter are hard-soldered to the top of each larger granule.
9Pendant reference 267 in Catalogue 1983 and reference 4 of Masov et al. (2005), page 120 (Fig. 2c) consists of one disc and one half-disc in cloisonné inlaid with turquoise and glass paste (identified by visual observation), bordered with granulated pyramids. The cloisons are 300-400 µm thick, and the encircling cloison about 2 mm large (Fig. 4). A 1500 µm rod shaped by hammering is bent as a hook and is joined to the back-plate of both elements, holding them together. An additional hook is joined to the back of the half-disc. One end of the hook was flattened while the other end was sharply cut straight.
Figure 3: The X-radiography shows that all the wires from pendant 272, as well as the sustaining rod that goes through the pendant, are hollow. The top of the rod is coiled around the suspension ring. Under the stereo microscope, the seams on the wires appear typical of strip-twisting.
Figure 3 : La radiographie à rayons X montre que tous les fils du pendentif 272 sont creux ainsi que la tige de soutien qui traverse le pendentif. La tige est ensuite enroulée autour de l’anneau de suspension. La morphologie de surface des filigranes sous loupe binoculaire montre les traces typiques des fils creux tordus.
10The outer borders of the setting boxes are surrounded by a row of granules of 1200 µm diameter topped by a tiny pyramid of four granules of 400 µm diameter on average (Fig. 4). The entire front part of the pendant was flattened, partially damaging a large number of granules.
Figure 4: Pendant 267: at the top, detail of the granules under the SEM; below, detail of the cloisonné under the stereomicroscope.
Figure 4 : Pendentif 267 : en haut, détail des granules au MEB et en bas, détail du cloisonné sous loupe binoculaire.
11Pendant earring reference 270 in Catalogue 1983 and reference 17 of Masov et al. 2005, page 124 (Fig. 2d) is typologically very similar to many objects from Tillya Tepe (Collectif, 2006). A long rod about 1000 µm in diameter, forged as a hook and serving as suspension loop and ear hook, is flattened to be joined to the central body (Fig. 5), consisting of one mango-shaped turquoise setting and one round-shaped garnet setting, both in cabochon, placed on the same back plate. Two gold strips, 2 mm large, form the box settings, which are surrounded by a row of granules about 1 mm in diameter (Fig. 5). A hammered disc is suspended with a 450 µm diameter coiled strip-twisted wire from the main body through a small ring made from a 750 µm diameter strip-twisted wire.
12Earring reference 271 in Catalogue 1983 and reference 8 of Masov et al. (2005), page 124 (Fig. 2e), is composed of a suspension hook and a complex pendant. The suspension hook consists of strip-twisted wire of uniform 750 µm circular section, each strip bent into a loop, interlocking facing opposite directions, resulting in a Gordian knot that supports a turquoise setting in cabochon surrounded by 700 µm diameter granules. These granules are half-melted, maybe because of a poor control of the joining process. On the back side of the hook, one set of 500 µm diameter strip-twisted wires (Fig. 6a) is joined on the inside of a cylindrical hinge 1000 µm large, while the other set of wires is inserted into the same cylinder, emerging for threading into the earlobe. A ring surrounded by one 850 µm diameter spooled wire is joined to the two dolphin-shaped handles decorated by chiselling and to the golden amphora’s neck. This type of wire is almost absent from Afghani gold work (Ogden, 2007).
Figure 6: Earring 271: (a) the X-radiography shows that the wires are hollow; (b) and (c) the different wires used in the decoration under the SEM.
Figure 6 : Pendant d’oreille 271 : (a) la radiographie X montre que les fils sont creux ; (b) et (c) au MEB les différents fils utilisés dans les motifs de décoration.
13A strip-twisted wire is wound around the upper part of the neck from where the amphora hangs. A coiled strip-twisted wire joins the amphora to the ring, functioning as a central rod. At its bottom hangs a flower-shaped bell on 100 µm thick gold foil, decorated with 500 µm diameter twisted and spooled wires (Oddy, 1977 definition), as illustrated in Figure 6b. Two suspension rings are joined to the external side of the dolphins’ tails, hanging a single loop-in-loop chain of 500 µm diameter strip-twisted wire; each chain is obtained by inserting four links into each other. A flower-shaped bell decorated with 500 µm diameter strip-twisted wires is suspended at the bottom of a coiled strip-twisted wire (Fig. 6c).
14The composition of the different parts of each jewellery item was determined by PIXE at the AGLAE accelerator of the C2RMF, with a 3 MeV external proton beam of 50 µm diameter and an intensity of 30-40 nA. The acquisition times ranged from 3 to 5 minutes. For the bigger parts, a region of 200 x 200 µm was scanned for homogenization. Two Si(Li) detectors were used to collect the X-rays emitted by the sample. One of the detectors is dedicated to the measurement of major elements, while the second detector, equipped with a 75 µm Cu filter to selectively decrease the intensity of the gold L-lines, can be used to identify minor and trace elements when higher acquisition times are used (Guerra, 2004; Guerra and Calligaro, 2004). The compositions of the different parts of the items are presented in Table 1. The limits of detection are on average 40 ppm for Cu, 300 ppm for Ag and 400 ppm for Au.
15The alloys used to fabricate pendant 267 have the following average composition: 82.0±1.2% Au, 14.7±1.0% Ag, 3.1±1.1% Cu. These alloys are of better quality than those used in the production of the other four objects, which have the following average composition: 75.2±2.5% Au, 22.7±2.6% Ag, 1.9±0.8% Cu. We must however note the slightly higher copper contents of some decorative elements of the disc of pendant 267 as compared to the elements of the half-disc.
16The compositions of pendant 272 and earring 271 are very similar; the average composition is the following: 76.3±2.1% Au, 21.8±2.3% Ag, 1.7±0.8% Cu. However, the granules of the dome plaque of pendant 272 have higher copper contents than the other elements of this item, and some elements have lower silver contents, such as the granules and the junction disc. The back plaques of earring 271 also present higher copper contents than the other elements of this item.
17Very few publications consider scientific analyses carried out on Kushan gold objects. However, the analysis of a few Kushan coins, of a gold pendant with garnets, and of a gold coin struck by Kanishka (Sachs et al., 2003) showed the use of gold alloys with compositions which are rather far from our results. The results obtained for the gold alloys used in the fabrication of the Kushan jewellery pieces analysed in this work are also distinct from the compositions found for several gold objects and decoration gold foils from a Xiongnu necropolis, which are separated in two main groups, with silver and copper contents of respectively 2% and < 1% for one group and ~10% and 1-2% for the other group (Guerra et al., 2005, Guerra and Calligaro, 2003). However, our results are very similar to the composition of the alloys used under Vasudeva II (after circa 261 AD) to strike concave dinars (Fussman, 2003).
Table 1: PIXE results obtained for the composition of the gold alloys used in the Kushan jewellery analysed (in %).
Tableau 1 : Résultats obtenus par PIXE pour la composition des alliages utilisés dans la fabrication de l’orfèvrerie kouchane (en %).
Voir PDF en annexe
See PDF below
18As previously noted, the pendant earring 270 is typologically very similar to several objects from Tillya Tepe. A very small set of objects from this necropolis was analysed at the C2RMF, and the results were published by Calligaro (2006). In contrast to the Kushan jewellery analysed in this work, all the analysed objects from Tillya Tepe are of very good quality. Among those objects, we draw attention to the pendant MK04.40.109, found in tomb V, and a gold appliqué MK04.40.113, found in tomb II (Collectif, 2006), dated to the 1st century AD, which contain on average 96% Au, 3% Ag and 1% Cu, and 97% Au, 1.8% Ag and 1.1% Cu, respectively. An equivalent composition was obtained for the ingot dated from the 4th-2nd century BC found in the Palace of Al Khaouïn, a town founded by Alexander the Great and situated close to the present day Tajikistan border.
19The analytical study of five jewellery items containing gold wires and granules from the collections of the National Museum of Antiquities of Tajikistan provides a first approach to the goldsmiths’ technological choices during the Kushan period. The non-destructive examination of the jewellery revealed the use of a large variety of techniques and decoration elements, sometimes very similar to those observed for the Afghani gold work of Tillya Tepe, such as the strip-twisted wires with seams running longitudinally along the wire, the absence of complex granulation patterns, the stone settings, the hammered elements, and so on.
20The non-destructive analysis of the objects by PIXE revealed the use of two distinct gold alloys, which contain on average about 15% Ag and 3% Cu for one type of alloy (pendant 267) and 19-26% Ag for the other (the other four objects), with copper contents that are either around 1% or around 2%. Data on particular elements, such as wires and granules, cannot be generalized, but, if we exclude pendant 272, we note that these elements have similar compositions to the other elements of one and the same item.
21Our elemental results are distinct from the compositions obtained by several authors for a few items and an ingot from Tillya Tepe, for several items produced by the Xiongnu, for one Kushan pendant, and for most of the coins struck by the Kushans. Only the concave dinars struck under Vasudeva II show a composition similar to our objects.
22We must note that the results obtained for these five objects cannot be generalized for Kushan jewellery, with technological traditions and skills that were influenced by Chinese, Iranian, Greek and Indian craftsmen. This combination of different traditions certainly also led to the use of many different alloys. In addition to this aspect, we must also bear in mind the diversity of the supplies in a region that benefited from contact with the trade routes between Orient and Occident.