Uniformity in Tophet Ceramics?

I The geological and geographical framework

Modern-day Tunisia is geographically divided into a northern region lying above the eastern part of the Algerian range of Atlas Mountains, a central divide of several low plateau and plains sloping towards Tunisia's eastern coast, and a southern region that is adjacent to the Saharan parts of Algeria and Libya.^[1] Tunisia’s northern and central regions are characterized by sedimentary rocks containing marine and lake sediments, clays, sandstones, limestone and argillaceous-sandy-fluviatile sediments underlie whereas the southern region mostly contains Mesozoic and Cenozoic marine sediments, dunes and alluvials with only a few areas containing sandstone, clays, limestones, calcareous marls and argillaceous-sandy-fluviatile sediments.^[2] Along the coast, marine sandstone deposited during the Middle Pleistocene contains pebbles of limestone and broken shell fragments were overlain by Late Pleistocene to recent continental alluvium. These younger sediments contain thick deposits of hard calcite.^[3] The main Carthaginian clay raw materials are situated in a geologically homogeneous area of sedimentary rocks, including sandstone and limestone beds, resulting in the availability of quartz-rich clays that could have supplied the raw materials necessary for the production of the urns and their lids. Ceramic fabrics derived from these raw materials therefore have a composition characterized by inclusions consisting mainly of carbonate features and quartz sand.^[4]

II Materials

The urns of the Carthaginian tophet precinct form a very specific assemblage and represent a morphological repertoire which identifies a series of discrete forms alongside various plates and bowls used as lids. The sample set presented and described here consists of 27 thin sections sampled from urns and lids excavated during the 1976–1979 ASOR Punic Project excavations (Table 1). The sampled assemblage includes material from early phases dated at around 750 to the fall of Carthage in 146 BCE. Within the sampled assemblage, morphology of the closed-form vessels includes wasp-waisted, stepped neck, everted-neck with vertical or horizontal handles, straight-necked chardon, and piriform shapes, next to a morphology of open-form vessels that includes umbilical saucers and generic bowls, as well as undiagnostic body sherds which do not carry detailed morphological information but are derived from known contexts and included because they complete the petrographic dataset. A full typological catalog of these vessel forms will accompany the forthcoming final excavation report, alongside detailed discussions about the dating of the different phases.

III Methods

Ceramic petrography has well-established optical microscopy techniques to identify and characterize different fabric groups within an assemblage.^[5] We distinguish between fabric groups on the basis of their mineralogical composition, the optical properties of their matrix, as well as the presence and size of pores and non-plastic inclusions.^[6] Over the past decades, petrographic analysis of thin sections has become a well-established technique for identifying and characterizing ancient ceramics from all over the world.^[7] We conducted thin section analysis and imaging on a Leica DM750P petrographic microscope at the Material Culture Laboratory at the Faculty of Archaeology of Leiden University.

Several research projects have provided evidence for local and regional pottery production at Carthage. For example, we have evidence for ceramic workshops and kilns in the southern part of the city, near the ancient coast below the Roman Cardo IX and for the presence of several kilns located in Dermech and Douimès from the Late Punic period.^[8] Moreover, north Africa has been the focus of several important petrographic studies that are especially useful for identifying the extensive production of amphora and tableware during the Roman period, as well as for differentiating other classes of pottery found at Carthage and surrounding regions.^[9] Although we have detailed macroscopic data for the Carthage assemblages,^[10] detailed and extensive petrographic characterizations of local Punic production have been restricted, at the moment, to just a few exploratory studies of various plain wares.^[11] This study presents a petrographic framework, building toward future a chemical compositional framework, for this particular set of Punic ceramics.

The main problem of conducting this type of analysis pertains to the rather generic compositional features of north African clay deposits, since tertiary deposits cover almost the entire coastal region from modern Algeria to Egypt. Nevertheless, ceramic production from northern Tunisia, especially in the Carthage-Nabeul region, differs from material produced in other regions, due to its enriched quartz in association with calcareous microfossils and/or limestone fragments.^[12] We sporadically encounter metamorphic and (mafic) volcanic fragments within the Tunisian assemblages, but not systematically. Although the material from the Carthage precinct is Punic, we deemed it important to assess these fabrics in light of later and better-known Roman amphora production.^[13]

IV Results and discussion

The petrographic analysis yielded the identification of six petrofabrics (CT A-F) on the basis of 27 samples (Fig. 1). Previously established fabric groups have mineralogical and textural characteristics that indicate discrete production processes and/or deliberate choices in the chaîne opératoire.^[14] Nevertheless, the characteristics of the matrices show only limited variation overall and each matrix has a quartz-rich and carbonate-rich component given the geological substrate. Carbonates are ubiquitous in arid regions, and across the central Mediterranean in general, which complicates the resolution of effective sourcing.^[15] The aplastic fraction shares this limited variability, and a common set of different mineral grains, often identified such as mono- and polycrystalline quartz, chert, feldspars and various opaque minerals (i. e., likely predominant iron oxides), while porosity can be described as micro- to macrovughs, voids and characteristic elongated pores.

Figure 1

Photomicrographs of the identified petrographic groups (images are in taken under crossed polarizers). 1) Petrofabric A: Sedimentary – (Sub)angular Quartz; 2) Petrofabric B1: High Fired – Rounded Quartz; 3) Petrofabric B2: High Fired – Rounded Quartz; 4) Petrofabric C: Ferruginous – Sedimentary; 5) Petrofabric D: Mixed Calcareous – Ferruginous; 6) Petrofabric E: Volcanic – Calcareous; 7) Petrofabric F: Quartz-Grog.

The groups presented here are based on the presence and composition of inclusions, different sizes, ratios of abundance of inclusions and roundness/sphericity of the grains in relation to ceramic matrix features. Matrix identifications relate to levels of a visible calcareous component, firing temperature and assessment of mixing of raw materials. Given the difficulty in discretely identifying provenance in north African fabrics, these groups are firstly aimed at the determination of systematic manufacturing sequences rather than positively identifying an actual clay outcrop. A detailed overview of the different identified petrofabrics can be evaluated in Table 2.

Group A: Sedimentary – (Sub)angular Quartz . Petrofabric A is primarily defined by an abundant presence of relatively small quartz and iron oxide grains (maximum sizes of 250µm but generally below 100µm) in an otherwise well levigated and homogenous fabric. We can detect sporadic small calcareous concretions and microfossils, in a generally dark reddish-brown matrix with limited optical activity under cross-polarized conditions.

Porosity is restricted on the whole to microporosity (<10µm) in association with small to medium vughs (< 150µm and ~500µm). The overall characteristics of this fabric closely resemble that of an amphora sample from Carthage Bir Messaouda (C1), dated to 360–340 BCE and identified as having a non-Carthaginian origin.^[16] A similar description appears regarding two groups attributed to early Carthage production sites – C1 dated 750 to 650 BCE and C2 dated from 650 to late sixth/early fifth century BCE.^[17] These pertain predominately to tableware morphologies, namely in the forms of plates and carinated cups.

Although the available set needs confirmation in terms of sample size and associated evidence, the presence of amphora and tablewares as well as burial urns would signify an interesting socio-economic connection that deserves further consideration.

Group B: High Fired – Rounded Quartz . This petrofabric has been split in two subgroups (B1 and B2), characterized by very similar properties but with an identifiably different matrix that necessitates the provision of subgroups for this fabric. Fabric B1 has abundant (sub)rounded eolian quartz as a main component in a low optical active and homogenous iron-rich matrix (under crossed polars). A frequent amount of relatively small (<150µm) iron oxide minerals are also commonly attested. The uniformity of the sintered matrix supports the identification of a high-fired fabric, and it clearly has a minor amount of cryptocrystalline calcareous content. Furthermore, this petrofabric can be identified by its typical porosity in the form of long and small, subparallel pores (generally up to 1-2 mm). Petrofabric B2 presents a high-fired, nearly vitrified version of the more common B1 variation. Small differences appear in carbonate content and presence of small mudstone particles.

This fabric has been identified systematically and constitutes a typical local Carthaginian fabric employed for the production of other classes of pottery, such as transport amphora.^[18] Samples studied by Maraoui Telmini consist mainly of amphora fragments, most of which are derived from the Punic kiln at Dermech, which supports for this fabric a designation as local.^[19] The presence of frequent subparallel porosity adds to its consistency and similarity. The material from this study derives from a Late Punic phase (first half of the second century BCE), implying a multifunctional continuation of the same raw material outcrops.

Group C: Ferruginous – Sedimentary Petrofabric. This fabric type has a characteristic feature of an optically active and semi-calcareous matrix with silty, ferruginous striations of varying thickness (10µm -100µm), as well as larger accumulation areas. The aplastic fraction has both subangular and subrounded fine to medium sized quartz grains, isotropic iron-oxides and a more limited presence of micritic concretions. Porosity characteristics includes both subparallel pores (although more restricted than group B1 and B2) and microporosity, as well as general voids.

Maraoui Telmini and Bouhlel describe “fine oxide impregnations in the matrix” in two instances from Utica (a jug and unguentarium in subgroup 7 from the necropolis), which may match the systematic fine ferruginous striations/bands that characterize this group.^[20] This preliminary association does not carry provenance significance beyond northern Tunisia but point to a raw materials and production processes designed for smaller plain wares rather than for transport ceramics, cooking wares or tablewares. Although the sample size is small, all piriform/straight neck urn types correspond to this particular petrofabric.

Group D: Mixed Calcareous – Ferruginous Petrofabric. Clear mixing bands combine non-calcareous and iron rich material with a high calcareous clay content. The matrix still has a relatively high optical activity which would suggest a slightly lower firing temperature as opposed to the previously described petrofabrics. Otherwise, a consistent high quartz content remains, in combination with cryptocrystalline calcareous inclusions, similar to previously identified Tunisian fabrics. Nevertheless, the extent and common practice of intentional mixing cannot be deduced from this rather small petrofabric grouping, but this combination either relates to a very different production sequence (i. e., intentional mixing of two clay sources) or the use of a very different clay outcrop (in case of a naturally mixed sediment).

Therefore, defining a potential origin for this particular petrofabric remains unclear. It petrographically remains consistent with northern Tunisian geology, but differs from common fabrics present at Carthage itself. Therefore, a regional production would be plausible given the uniformity in inclusions composition. One notes that limestone-based outcrops along the central Tunisian coast possess higher amounts of carbonate materials. Moreover, modern wares produced at Moknine still have lime added deliberately to their recipes,^[21] possibly a consistent practice for this type.

Group E: Volcanic – Calcareous Petrofabric . This petrofabric group contains several volcanic (basaltic) rock fragments in an otherwise very fine-grained calcareous matrix. Other inclusions present are mono- and polyquartz as well as small grog grains (<200 µm). This area of northern Tunisia lacks this kind of volcanic (or metamorphic) rock and mineral inclusions. The clear presence of these inclusions in our samples, as well as their systematically different matrix properties, indicate this petrofabric is not consistent with northern Tunisia but may serve as a potential petrographic marker for imported vessels. Both Pantelleria, Sicily and surrounding regions commonly have these inclusions and would serve as the likely source of raw materials for these particular urns and lids. Nevertheless, since Pantellerian wares are generally characterized by their abundant feldspar mineralogy, predominantly anorthoclase, this seems at least for now to be a less likely origin for these particular vessels.^[22]

Group F: Quartz-Grog . Lid sample 5578A and an undetermined base fragment 5606 contain large grog inclusions (up to 2 mm) and have a coarse sand size fraction (absent from the rest of our tophet materials), with an iron-rich, non-calcareous matrix. The aplastic fraction contains a minor cryptocrystalline calcareous content, few bioclasts and both quartz and quartzite. Fabrics enriched in grog have been described by Maraoui Telmini and Bouhlel from the Besbassia kiln in Utica,^[23] but confirming this connection will require direct comparison of thin sections.

Petrofabrics Overview . A preliminary observation of these thin sections reveals a predominance of iron-rich fine-grained fabrics. Although carbonate presence varies, significant calcareous content appears in the form of (cryptocrystalline) limestone, or fossiliferous materials, and remains consistent overall with the commonly attested ceramic assemblage of Carthage. While this calcareous occurrence seems quite restricted in relation to urn and lid manufacture, the consistent presence and morphology of the quartz grains proves a key attribute of the main Carthage production groups A and B1–2,^[24] while groups C and D can be situated with the wider northern Tunisian/Carthage-Nabeul production region. Group E should be regarded as an import based on its particular volcanic inclusions and different matrix properties, related to another location in north Africa or to the islands, such as Sicily and Sardinia. Petrofabric F, enriched in grog, might align with similarly enriched samples identified at Utica.

V Conclusions

The microscopic analysis of the ceramic urns and lids clearly yields the identification of several production sites which supplied the ceramics used during the activities at the tophet precinct. While an elaborate ceramic production was well-established at Carthage, we can exclude a systematic and dedicated supplier for this specific precinct. In contrast, the urns and their lids seem to be aligned with other ceramic production sites (e. g., the local plain wares and transport ceramics) attributed to the local Carthage répertoire. Punic ceramic production from other nearby areas has not yet provided direct parallels, despite close resemblances due to the uniform geological substrates. Nevertheless, as established by several other studies, a network of regional workshops, utilizing discrete clay resources or at least practicing different technological procedures, had developed in the Punic period. While the majority of urns and lids clearly have a local signature, connecting to established ceramic manufacturing and directly pointing to production at Carthage, just a few samples suggest some vessels travelled also from further afield. The ceramics for this precinct were assembled from various production sites independent of time and morphology. While the sampled material is regional in scope rather than Mediterranean, distinct production centers may be established after further investigation. Whether these particular urns and lids were put to other uses prior to being deposited or were made specifically for use in the tophet precinct, must remain open questions for now, but will provide an important avenue for future research.