- 1 Gorton, 1996. Masson, forthcoming.
- 2 Webb, 2013-2015a.; Webb, 2013-2015b.
- 3 Masson, 2013-2015.
- 4 Pierrat-Bonnefois, Bouquillon and Coulié, 2014; Caubet and Pierrat-Bonnefois, 2005; Busz and Gercke (...)
1The city and trading port of Naukratis in Egypt, inhabited by both Greeks and Egyptians, was an important faience production centre, especially between 600 and 550 BC. Scarabs and other amulets manufactured in the so-called “Scarab Factory”, excavated by W.M.F. Petrie in 1885, were exported throughout the Mediterranean world1. It is likely that other faience products were also made in Naukratis and distributed widely, for example figurines of musicians which can be dated from separate evidence down to 550 BC2 and New Year’s flasks3. In addition to some typical Egyptian motifs, many scarabs and scaraboids feature a hybrid Greek-Egyptian or Near Eastern iconography. This has raised the question of the ethnic identity of the craftsmen involved in this factory: local Egyptians, Greek settlers or Phoenician craftsmen? Although faience technology was already known in Egypt since the Predynastic Period and had existed in Bronze Age Crete and possibly also Mycenaean Greece, it had disappeared along with those cultures at the end of the Bronze Age and it is only during the first half of the first millennium BC that the technology was reintroduced to the Greek world. The Greek island of Rhodes is considered to be one of the earliest centres producing faience at least partly in a Greek style, followed later by Naukratis. The routes and agents of this technology transfer remain little understood, in particular the respective role played by Egypt and the Levant/Phoenicia, which both have strong traditions of local faience production4.
2The collaborative study between the Musée du Louvre, the Centre de Recherche et de Restauration des Musées de France (C2RMF), the British Museum and the Ashmolean Museum was established as part of a wider programme coordinated by Geneviève Pierrat-Bonnefois. This project aims to shed new light on the production of faience and trade networks, cultural contact and technology transfer in the ancient Mediterranean and Egypt during the first half of the 1st millennium BC.
3A large and representative number of faience objects from Naukratis and Rhodes were selected for analysis, comprising twenty-six objects from Naukratis, seven from Kamiros on Rhodes and one of unknown origin which was excavated at either Naukratis or Kamiros in the British Museum, one from Naukratis in the Ashmolean Museum and eight from Naukratis and one from Kamiros in the Musée du Louvre. They included faience vessels, scarabs, scaraboids and figurines, as well as elements related to the manufacture of these objects such as wasters. The selected figurines comprised all the main categories: musicians (such as lyre, flute or drum players), naked women and animals, mainly falcon, ram and lion. A few Egyptian style amulets were included in the group, such as Pataikos amulets and Bes heads.
Fig. 1. New Year’s flask neck BM 1885,1101.29
© Trustees of the British Museum.
- 5 Technical information about the equipment and illustrations of its use can be found in: Salomon, Dr (...)
4PIXE analysis was performed with the AGLAE 2 MV ion accelerator of the C2RMF located in the basement of the Louvre5. The ion beams are extracted to the atmosphere through a thin window. PIXE is performed with 3 MeV protons. A helium flow is maintained on the beam paths (particles, X-rays) to minimize energy losses and scattering. PIXE is used to obtain an average chemical composition of materials assumed to be homogeneous in the volume covered by the X-ray emission. The GUPIX code was used to determine major and trace oxide/element concentrations.
5Compositional data was acquired for 44 objects (see Table 1). More than one point/area on objects with multiple colours of glaze was analysed, resulting in a dataset of 57 analyses. These analyses were carried out on the internal body of damaged objects as well as the outer glaze layer. Only the results for glazes will be reported in this article.
Table 1. PIXE-PIGE results for all glaze areas analysed from Naukratis and Kamiros (in bold) organised by glaze colour
Results are normalised to 100%. Where the original colour is difficult to define, the compositional characteristics have been used to assign a colour to the object and a ‘?’ used to note this. Where P2O5, SO3, Cl and CaO values are higher than 5 wt% results for these objects should be considered semi-quantitative.
- 6 Newton and Davison, 1989, p. 136.
- 7 Tite, Maniatis, Kavoussanaki, Panagiotaki, Shortland and Kirk, 2009.
6As with any form of surface analysis, there are some issues with the data. Faience, like most vitreous materials, deteriorates over time and the composition of the surface is often not representative of the original composition of the glaze. This process normally results in a loss of alkali components (soda [Na2O] and potash [K2O])6. It can also cause alterations in the colour of the glaze7.
- 8 See: Tite, Freestone and Bimson, 1983; Tite, Freestone and Bimson, 1987; Tite and Shortland, 2003; (...)
- 9 Pollard and Heron, 2008, p. 126-128; Strahan, 1991.
7In the case of the objects analysed here, there is also a problem with contamination. The majority of the analyses have elevated levels of calcium, sulphur, phosphorus or chlorine when compared with previously published analyses8. This suggests that the objects have undergone a process of post-depositional alteration and are contaminated with sulphates, carbonates, chlorides and phosphates9. This will not only cause issues with the calcium, sulphur, phosphorus or chlorine levels reported in this work, but also the levels of all other components will be reported below their actual amounts present below this contaminated surface layer. Where levels of these components are above 5 wt%, the data for these area analyses should be considered semi-quantitative (see Table 1).
8Despite these problems, the use of non-destructive PIXE analysis at AGLAE was indispensable, since the very small size of many of the objects essentially made sampling impossible. The results set out below moreover confirm the method’s efficacy.
9On first study of the results, the glazes of the faience objects from Naukratis and Kamiros appear to have very similar chemical characteristics. Oxides which are components of the basic raw materials (such as potash and alumina, see fig. 2) were found to be variable at both sites, but there was a high degree of similarity between sites.
10No yellow glazed objects from Rhodes were selected for analysis during this project. The yellow glazes analysed on objects from Naukratis were all found to contain significant levels of lead (Pb) and antimony (Sb). The presence of lead antimonate (Pb2Sb2O7) crystals within a glass or glaze imparts opacity and a yellow colouration. There is a weak correlation (PbO/Sb2O3 ratio has a fairly small range, between 1.5 and 3.5) between the quantities of lead and antimony present in these objects, apart from one scarab (BM 1886,0401.1625), suggesting that they may have been added together as parts of a single raw material. However, further analysis would be needed to confirm this. The one yellow scarab which does not fit this pattern (BM 1886,0401.1625) contains exceptionally high antimony oxide levels and has a PbO/Sb2O3 ratio of 0.8. This object was probably produced using a different recipe and/or antimony-rich raw material.
- 10 Duran, Castaing, Lehuédé and Bouquillon, 2012.
11A further number of the yellow scarabs have iron and tin levels which varied significantly from the other yellow objects. One of the yellow scarabs analysed was found to contain high iron levels (BM 1886,0401.1608) and three others contain a significant level of tin oxide (Louvre E 8056 bis 12, Louvre E 8056 bis 5 and BM 1886,0401.1625) (Table 1). The presence of tin and iron in lead-antimonate particles is common and is probably linked to the heterogeneity of the raw materials used10.
12The green/blue-green glazes are the most common in the analysed assemblage. This colour is sometimes homogenous on the best preserved artefacts, but often only coloured spots remain on a powdery faience body (see fig. 1). Differences in the results of the chemical analyses have allowed us to propose two main compositional types for the objects from Naukratis.
13The first type is characterised by glazes which contain significant levels of antimony and lead. The colour of these objects is produced by mixing copper oxides, and associated varying levels of iron oxide, with lead antimonate. The hue is a function of the relative quantities of each component. This group consists of four objects from the British Museum (three scarabs BM 1886,0401.1621, 1886,0401.1626 and 1886,0401.1660, one New Year’s flask BM 1886,0401.1589) and two scarabs from the Louvre collections (Louvre E8056 bis 9 and 11). One further object with very low antimony oxide levels (New Year’s flask, BM 1885,1101.29, fig. 1) may be included in this group. This low antimony, potentially correlated lead-antimony object is also the only object in the entire assemblage which contains high levels of tin (c. 2wt% SnO2).
- 11 Kaczmarczyk and Hedges, 1983.
14As with the yellow glazes, the correlation between lead and antimony seen in a number of the results suggests that they may also have been added as a single ingredient to these green/blue-green objects. Objects of these colours previously analysed by Kaczmarczyk and Hedges11 also show a correlation between lead and antimony. The fact that this correlation was also found by other researchers strongly suggests that a single ingredient may be responsible for the presence of these components in a significant number of the green/blue-green coloured objects.
15The second group consists of 10 objects. Lead is present as a minor element (average 0.2 wt% PbO) and very little, if any, antimony was detected in these objects (fig. 3). Their colour is only caused by the presence of copper and/or iron oxides. The presence of copper and lack of lead antimonate in these glazes is likely to have given them a turquoise appearance when produced. The colour could have become greener through weathering processes during burial or, as for BM 1886,0401.1641 (couchant lion) where iron was present in significant quantities, a green colouration would have always been present.
16The three green/blue-green glazed objects from Kamiros have varying lead and antimony levels and also fall into two groups (fig. 2, Table 1). Two objects have fairly similar lead and antimony levels (c. 0.75 wt% Sb2O5 and 3 wt% PbO) and are very similar in appearance (scarabs BM 1861,0425.12 and 1861,0425.13). They are relatively free from weathering and have the strongest blue-green colour of any of the objects analysed in this study. These two objects share extremely similar chemical characteristics and were therefore made from an almost identical set of ingredients and recipe. The third object has a different form (ram, BM 1861,1024.22) and contains very little lead or antimony, similar to the majority of objects analysed from Naukratis.
Fig. 2. Plot illustrating the relationship between potassium oxide (K2O) and aluminium oxide (Al2O3) levels in the glazes of the faience objects analysed from Naukratis and Kamiros
Fig. 3. Plot illustrating the relationship between lead oxide (PbO) and antimony oxide (Sb2O3) levels in the yellow and green/blue-green glazes of the faience objects analysed from Naukratis and Kamiros
The data for one yellow object (scarab, BM 1886,0401.1625) is not included in this plot. This object has exceptionally high antimony oxide levels (>10 wt% Sb2O3) and is discussed above.
- 12 Shortland, Tite and Ewart, 2006; Kaczmarczyk and Hedges, 1983.
17Four objects are characterised by a more or less obvious blue colour. They all contain significant levels of cobalt (>0.07 wt% CoO). Three were excavated in Kamiros; two very similar scaraboids (BM 1861,0425.20 and 1861,0425.21) and another scarab (Louvre E 3899)(fig. 4a-c). In these objects, the presence of cobalt is associated with elevated levels of other elements (iron, nickel, manganese, aluminium and magnesium). This association is common in deep-blue glasses and glazes coloured using cobaltiferous alums from the Western Oases of Egypt12. The variations in cobalt concentrations are probably due to different mixtures of colorants, rather than changes in the cobalt source used. Copper also seems to have been added separately, sometimes in very high quantities, to alter the hue and obtain the brightest blue colour.
Fig. 4. Cobalt-containing blue glazed objects
a) Scarab with image of a lion (Louvre E3899 width 0.88 cm) © Musée du Louvre/Christian Décamps. b) Scaraboid with image of sphinx (BM 1861,0425.20 width 1.27 cm), c) Scaraboid with image of wild goat or ibex (BM 1861,0425.21 width 1.27 cm), d) Falcon waster (BM 2013,5012.11 height 2.20 cm).
© Trustees of the British Museum.
18The findspot of the fourth object containing cobalt (falcon waster, BM 2013,5012.11, fig. 4d) is uncertain, but it is known to come from either Naukratis or Kamiros. This object contains very low levels of nickel along with cobalt, and this may have been an impurity added along with the cobalt source used in its production, similar to that found in the three other blue objects from Kamiros.
19The dark-coloured decoration (often described as black or brown) on many of the objects is produced by the addition of large quantities of manganese, normally in combination with iron. Two main compositional types can be proposed for glazes of this colour.
- 13 Kaczmarczyk and Hedges, 1983.
20The first group includes the largest number of objects, 10 pieces, which are all from Naukratis. In this group, the black glazes are coloured with a mixture of manganese and iron. In a few cases, barium is associated with manganese. This suggests that different manganese-containing ores have been used. Where significant levels of barium were detected, romanechite or psilomelane can be proposed as the manganese source. Where very little or no barium was detected, pyrolusite-rich deposits were probably used. Similar differences have already been observed by Kaczmarczyk and Hedges13.
- 14 Pierrat-Bonnefois, Becq, Bouquillon and Lehuédé, 2015.
21The second group can be very clearly differentiated from the first. All of the objects in this group contain low levels of cobalt and nickel, as well as manganese and iron (fig. 5). This may be an intentional addition that plays a role in the final appearance of the glaze and enhances the dark colour. Such a mixture is detected in the dark glazes of the three objects analysed from Kamiros and only on one (a musician figure, BM 1886,0401.1330) found at Naukratis. This recipe seems to be strongly linked to Kamiros. Indeed, previous analyses of black glazes on four figurines from Kamiros (Louvre NIII 2407, AM 403, AM 404 and MN 2416) show the presence of cobalt and nickel, associated with manganese14.
Fig. 5. Plot illustrating the relationship between nickel oxide (NiO) and cobalt oxide (CoO) levels in the dark-coloured glazes of the faience objects analysed from Naukratis and Kamiros
- 15 Pierrat-Bonnefois, Becq, Bouquillon and Lehuédé, 2015.
22The preliminary study of the data has revealed many interesting similarities and differences with, and between, the chemical composition of faience objects from Naukratis and Rhodes. Many of these patterns correspond with those previously found in the analysis of other faience items from the collections of the Louvre15.
23It is clear that there were similarities in the raw materials and recipes used to produce the faience found at both of these sites. However, some differences can be seen in the materials used to colour these objects. At this stage, the differences and correlations in lead and antimony levels in yellow and green/blue-green glazes, and identification of differences in the use of cobalt to produce the dark-coloured glazes, seem to provide the greatest hope in showing differences between the two production areas.
24The correlation found between lead and antimony levels in some yellow and green/blue-green scarabs and New Year’s flasks from Naukratis suggests that a single material containing both of these ingredients was used in their production. In the future, this may provide a means of characterising some of the faience from Naukratis. A comparison with Rhodian faience (small figurines and vases, but no scarabs or ram amulets) found in excavations performed by Salzmann and Biliotti at Kamiros show similar trends for green/blue-green faience. The picture is not so clear for yellow faience, as the assemblage from the British Museum analysed in the present study does not include any yellow faience object from Rhodes.
25The results for blue and dark-coloured faience items suggest that a nickel-containing cobalt source may have been used in their production at Rhodes, and not at Naukratis. Further analysis of a greater number of objects from both sites will be necessary to confirm this tentative pattern. If it can indeed be confirmed, it has the potential to be used as a provenancing tool.
- 16 See: Tite, Freestone and Bimson, 1983; Tite, Freestone and Bimson, 1987; Tite, and Shortland, 2003; (...)
26It must also be borne in mind that production in Naukratis was confined to a shorter period than production on Rhodes, which ranged from 650 through to 525 BC. Further analysis and rigorous comparison with previously published studies of ancient faience production16 will be necessary before strong conclusions can be drawn.
AGLAE beam time was made possible under the FIXLAB transnational access programme of the CHARISMA project. Financial support by the Access to Research Infrastructures activity in the 7th Framework Programme of the EU (Grant Agreement n. 228330) and the help of the AGLAE team is gratefully acknowledged.
Flux RSS
