Undecorated Roman-Period Roof Tiles – An Old Material Providing New Results

1 Introduction

Earthen architecture comes in many different forms, and one of the most common materials is fired clay in the form of bricks and tiles. The expansion of the use of this material in Europe began in the Roman Imperial period, although fired building ceramics themselves have a much longer (pre)history (e.g. Sapirstein, 2016). Decorated and undecorated ceramic roof tiles and their fragments are one of the most enduring and numerous remains of the earthen architecture of the Roman period to survive to us from many parts of the ancient world, especially from the Mediterranean region. Clay as raw material allowed for a great variety of forms and shapes for ceramic architectural elements, and firing made them extremely durable. Raw materials for their manufacture were commonly available, and bricks and tiles were relatively affordable to manufacture. All this allowed the development of a wide variety of clay roof tile types during Antiquity, from plain and utilitarian to highly specialised decorative forms.

It is therefore slightly surprising that research into ancient ceramic roof tiles has not been more extensive. Although specific phenomena, such as tile stamps, decorative terracottas, and the early Archaic tiles, have been the focus of studies (e.g. De Domenico, 2019; Hübner, 1997; Sapirstein, 2009), plain, undecorated tiles are only infrequently studied. Due to this sparsity in specialist studies and published excavation or survey records, we have huge gaps in our knowledge regarding this material. The most detrimental gap is a lack of sufficiently fine-grained tile typology to easily situate and compare any tile material found. Typologies would help us with dating; they would also help answer questions related to trade, economic conditions, and workshop structure (Lancaster, 2015). In addition to the absence of an accepted typology, fundamental questions related to e.g. the distribution of types, the production of plain tiles, and crafting processes are lacking. This lack is especially keenly felt in the eastern parts of the Roman Empire, where research has been impacted by the strong focus on the early decorative tile systems and the limited amount of stamping in the eastern part of the Empire (Hamari, 2019; see also Hübner, 1997; Winter, 1993).

Such gaps in a type of object that is available in abundance, but simple in form, might suggest that there is little to be gained from a more advanced study, which this study argues against. This work will in particular outline the methodology that can be applied to the study of plain tiles, used in the author’s recent thesis (Hamari, 2019) concerning Roman-period ceramic roof tiles in the Eastern Mediterranean. This summary review presented here will discuss the different ways we can methodologically approach undecorated roof tiles as material, extending from traditional typological studies to technical compositional analyses. The original research, published in a series of articles (Hamari, 2008, 2017a,b; Hamari, Tsiafaki, Kazakis, & Tsirliganis, 2019) and summarised in the thesis summary (Hamari, 2019), covers the full details of the materials included and the justifications for the conclusions made. However, a short introduction to the assemblages used as the basis of the research, as well as a more detailed discussion on the methodology applied and the wider conclusions made in the original research, is presented here. The implication is that considerable gains in further knowledge lie in the study of plain roof tiles, made possible and justify a better documentation of the material in the future.

2 Looking at Roman-Period Plain Roof Tile Material

The motivation for such research and the questions raised thereby were made possible by the availability of some well-defined assemblages of plain roof tiles from across the Eastern Mediterranean from recent excavations. The characteristics of these primary assemblages (Paliambela in Greece, Petra in Jordan) are explained in more detail in Sections 3 and 4. Chronologically, they cover a period from the first to the fifth centuries CE, coinciding with the Roman dominance in the area. Spatially, they extend from Roman Greece to Roman Near East (see Hamari, 2019 for details). The published fragments amount to ca. 900 items in total. However, in all of the sites studied, all tile finds were checked and sorted, representing a larger finds base from which to draw data, with only a representative sample retained and published. In addition, a comprehensive desk study was made of any published roof tile finds from the whole area and period covered. Although the search for comparisons does not claim completeness, a fair amount of the published finds available, specifically from Greece and Jordan but extending to modern Turkey, Cyprus, and Syria as well from ca. 30 different sites, were covered for drawing the conclusions presented here.

The common factor in the primary assemblages was that they provided a first-hand opportunity to study excavated plain tile assemblages in a coordinated manner: all three assemblages consist of excavation finds that could be documented in situ or studied contextually as part of an excavated assemblage. Another valuable factor was that they provided a snapshot across several different types of roof tiles used in the Eastern Mediterranean in the Roman period in different areas, leading to the emergence of the trans-regional variation in types as the main research question. Although a general understanding of the development of roof tile types over time in this wide area exists, it is sufficiently accurate only for the early, Archaic, and Classical decorated tiles, whereas the later periods are only generally covered. The previous studies on tiles in this region with a typological focus are valuable but few, and provide information mostly on a specific site, region, or subtype (see e.g. Mills, 2013; Vriezen & Mulder, 1997).

In terms of the typology of roof tiles in the Eastern part of the Mediterranean, the types in use from Archaic through Classical and Hellenistic periods can be placed within the major categories of Laconian, Corinthian, and Hybrid systems, each with their own combination of pan and cover tiles (Figure 1). System in this case means a combination of a specific pan tile and a corresponding cover tile to cover the joints between the pan tiles. These were both curved in the case of the Laconian system, whereas the Corinthian system used a combination of a flat pan tile and a hipped cover tile. The Hybrid system was a mix of these two, with a flat pan tile and a semi-circular cover tile (see Hamari, 2019, pp. 18–23 for more detailed discussion and further references). These systems and their elements were already identified early in archaeological research (Hübner, 1997) and remain generally valid categories today, emerging originally in the seventh and sixth centuries BCE. There is some development over time in specific features, such as size or surface treatment, yet the systems remain essentially recognisable over extended periods of time and in different areas, extending also to the Roman period. This is because ancient roof tile manufacturers (or those commissioning materials) chose to adhere to these specific combinations of forms – clearly a traditional, even conservative craft relying on well-tested prototypes. This slow change combined with an adherence to generic forms means that we can follow the development of the types for a considerable period of time and over significant societal changes.

Figure 1

The Archaic Greek and related roofing systems (Laconian, Corinthian, and Hybrid systems), after Wikander (1988, Figure 6), following Åkerström (1966). Note that the Hybrid system is sometimes called the Sicilian; however, this term is not well suited and was replaced already by Åkerström with “Hybrid.”

Based on the finds available for my research and looking across the board to comparative materials, it became clear early on in my study that these broad categories, although generally relevant, were inadequate when describing the change and variability evident in the material. There was clearly more typological variation in specifics within the major systems (Laconian-style, Corinthian-style, and Hybrid) than previously indicated. A central argument of my thesis, which I recap here, is therefore that looking at the changes within the major systems in more detail, a more nuanced picture can be built on the temporal and stylistic variation within the categories, specifically valid on a regional level. This required a new approach to the material, using old methods which however had not been applied in such detail before to this particular material. This study will examine especially the methodological approaches used, and summarise the conclusions and the new results that could be obtained from these old methods and materials.

3 Roof Tiles from Roman-Period Greece

Of the assemblages involved, the Early Christian church site of Paliambela, Arethousa in N Greece represents examples of the types of tiles that were in use during the major building boom of the fifth century CE in the Greek area. This small site in the hinterlands of Thessaloniki is similar to a number of Early Christian church sites in the area. The site was excavated by the Finnish Institute in Athens between 1999 and 2004 (Karivieri, 2017). The ceramic building material from the site consists of pan and cover roof tiles, bricks, and other miscellaneous materials, dated to the Late Roman and Early Byzantine periods (third century CE to sixth century CE), published in detail in Hamari (2017a). The catalogued and published finds include ca. 300 roof tile fragments from 47 different tiles, including five fragmentary but complete tiles (Figure 2).

Figure 2

Laconian-style roof tiles of the destruction layer in the atrium area, waiting for documentation (Squares J96–J98) in Paliambela. Photo Pirjo Hamari, 2000.

The roof tiles from the church are so-called Laconian-style tiles with curved pan tiles and semi-circular cover tiles (Figure 3). The majority of these come from a unified destruction layer covering the whole area, consisting mainly of roof tiles, stones, and earth, possibly a result of a major earthquake causing the roof to fall in a single event. This destruction layer was soon covered with a layer of soil/earth, effectively sealing the layers underneath. Such excellent preservation of the material is noteworthy even in Greek contexts, as tiles more usually are found in more fragmentary conditions, especially in surveys, and therefore provides an excellent view on the Laconian-style roof tiles in their Late Roman – Early Byzantine format.

Figure 3

A Laconian-style pan roof tile from Paliambela, Cat. no. 17. Drawing Hamari, 2017a.

These Laconian-style pan tiles were moderately curving in profile, with a smooth concave upper surface and a slightly rougher convex lower surface, with an average size of 450 mm × 850 mm × 25 mm. The curve of the tile profile becomes more pronounced near the long edges, and sometimes the centre profile of the tile is nearly flat; however, the overall shape is not tapering. The long sides have slightly thickened and profiled flanges in various shapes protruding outwards (Hamari, 2017a, p. 44). The cover tiles are semi-circular and slightly tapering. A particular feature in these tiles are the abundant finger-line decorations known as signatures (Hamari, 2017a; Hamari et al., 2019). These finger-line markings become very common on the surfaces in the Laconian-style tiles especially towards the Late Roman period, but they are known already from tiles of the Hellenistic and Roman periods (Hamari, 2017a, p. 48). Based on a comparative analysis of similar published finds, the Paliambela tiles are generally representative of its type in this period and region, yet particular to this site or area, as discussed in Section 4.

4 Tiles in the Roman and Byzantine Near East

The two tile assemblages from Jordan (Monastery of Aaron, Hamari, 2008, and House IV on Ez Zantur in Petra city, Hamari, 2017b), in contrast, come from secondary deposits and are generally more fragmentary than the tiles from Paliambela. Although not as well-preserved or “closed” as the Paliambela material, their value is to be found elsewhere.

The tile material from the Nabataean-Roman town villa of Ez Zantur IV is interesting for several reasons. One is related to the general rarity of finding tiles in any assemblages in this area outside the direct influence of the Hellenistic kingdoms; all finds are of value in this capacity. This material is the first comprehensively retained and documented ceramic roof tile material from the city of Petra, excavated by the University of Basel in 1996–2001. The second is related to the good documentation policy of the excavation; even a decade after the excavation, a comprehensive picture of the tile assemblage could be reconstructed, including relative frequencies and types present. Most interesting, however, is the typological variety of the fragments and their dating, providing information about early roof tile use in Petra in general; the fragments do not derive from tile roofs of the house but instead are a collection of recycled tiles from other (public) structures in Petra city in secondary use. The tile material from the house of Ez Zantur is datable to the first century CE, and possibly even earlier (Hamari, 2017b, p. 93). Although this material is not yet published in full, the draft catalogue includes 170 entries with ca. 500 roof tile fragments in total. A small selection of these is already presented in Hamari (2017b).

The other example from Petra comes from the Monastery of Aaron, a monastic complex which had a long lifetime from the first to seventh century CE. Between 1997 and 2007, an international team led by the University of Helsinki conducted research at the Jabal an-Nabi Harun (“Mountain of the Prophet Aaron” in Arabic), which is located about 5 km from the centre of the ruined city of Petra. In terms of roof tiles, the most likely period of their manufacture and primary use is in the fifth century CE (Hamari, 2008). Originally used in roofing the monastery church, the material was later used as secondary construction material in the subsequent construction phases of the complex. Despite the secondary find contexts and fragmentary condition of the material, the assemblage has substantial value for research. It adds to the typological series of roof tiles in the area in the same manner as the recycled material from Ez Zantur does, only from a later period, as well as providing evidence on the production and provenance of tiles in this area. The catalogued and published finds include 110 tile fragments (Hamari, 2008, p. 377).

Both of these assemblages together provide a rare picture on the typology of roof tiles used in the area, and also provide a partial time series of types from the first to the fifth century CE. Details of the materials have been published elsewhere (Hamari, 2008, 2017b), but as a summary, although the assemblages show considerable variation between different subtypes, they belong to the Hybrid category of roof tiles: flat pan tiles with rectangular flanges combined with semi-circular cover tiles (Figure 4). A fortunately preserved and restored example of one type (Ez Zantur I, Hamari, 2017b, p. 91), clearly common in the city, is visible in the heated winter triclinium floor of the house (Figure 5). The Hybrid system, as the name states, was a system combining a flat pan tile with rectangular flanges (deriving from the Corinthian system) with semi-circular cover tiles (of the Laconian system). Adopted around 600 BCE in Asia Minor via influences from mainland Greece, this system becomes the preferred tile system in the eastern end of the Mediterranean (Hamari, 2019, p. 58; see also Åkerström, 1966; Winter, 1993). Based on available evidence, this type gradually spreads into Near East during the Classical and Hellenistic periods, forming part of the Hellenistic-themed architectural repertoires (Hamari, 2017b). However, although recognisably of the Hybrid “family,” the roof tiles from Petra show at least two very different versions of this type, emphasising that these types were in slow but constant change, and that regionally developed types could be very specific in style.

Figure 4

Some examples of fifth century CE roof tile types from the Monastery of Aaron site. Drawing Hamari, 2008.

Figure 5

In situ roof tiles used in the floor construction of the winter triclinium (Room 14) with a hypocaust in House IV at Ez Zantur, Petra. Photo: Pirjo Hamari, 2011.

Due to their nature as secondary deposits, the fragments in the assemblages in Petra are not as homogenous as the Paliambela material. Instead, they demonstrate a very varied range of types, emphasising that if you collect examples that span a wider period of time, even on one site, you get a lot of variation in the subtypes, even when they all belong to the wider Hybrid category. A similar picture, although preliminary due to the scarcity of comparative finds, is presented from other sites in the area, such as Gerasa (Ebeling, 2020), Gadara (Vriezen & Mulder, 1997), and Berytos (Mills, 2013).

Looking across these assemblages and related research, in terms of typology and implications to it, it was clear that in each of the sites under study, the tiles had a recognisable general form but with a lot of sub-variation, and none looked exactly like the western Imperial tegulae-type roof tile. These, although generally Hybrid in appearance, had developed from Archaic originals into quite standardised independent types with clear cut-outs on corners, flanges formed with frames, smaller in size, and differing in form (Figure 6; e.g. by Clément, 2013; Shepherd, 2007, 2015, Winter, 2009, Wikander, 2017, Warry, 2006; see also Hamari, 2019, p. 67 for discussion). In particular, their standard defining features in the Roman Imperial period are the different types of cut-outs in the corners of the pan tiles; their application as a basis for typologies in the plain tiles of the Imperial West has been specifically studied by Shepherd (2015) and Warry (2006).

Figure 6

The two types of plain roof tiles from Italy with differing cutaways, Shepherd, 2015, Figure 1 (used with permission).

In contrast, in the East we find different tile systems in use, with a wider internal variation in subtypes. Clearly, in terms of typology, in the Roman period, we are facing another reality in roof tile types in the Eastern Mediterranean than in the West.

5 Emerging Research Questions

In dealing with the Paliambela material, I had already quickly come across how little published material was available on comparative finds. The common approach in object studies is to situate any types against other existing examples or find their place in general typologies, and from that, draw conclusions on dating, function, movement of material, ideas, and people. This is a standard approach in pottery studies; however, it was evident that this approach was not available for tiles in necessary detail. For roof tiles, we have gaps in our knowledge in general studies to some extent, but more specifically in specialist studies and published excavation or survey material, despite some outstanding examples. The most detrimental gap is a lack of sufficiently fine-grained tile typology to easily situate and compare any tile material found (Lancaster, 2015).

Such gaps in research concerning a finds category that is available in abundance but simple in form might suggest that there is little to be gained from a more advanced study. Mentions of tile finds included in reports repeat words such as “standard,” “unremarkable,” “simple,” and “usual” – tiles are regarded as one of the “less impressive waste materials” (Witte, 2012). Therefore, one meta-level question addressed in the research was to assess whether a detailed study of plain tile assemblages would give added value for the archaeological interpretation of past societies.

A full analysis of the available evidence with references, both from the sites presented and from available published research, is included in my thesis (Hamari, 2019) and related articles (Hamari, 2008, 2017a,b; Hamari et al., 2019). This work focuses on the methodology applied to the study of plain roof tiles in support of future research on the topic and only briefly summarises the main conclusions.

6 Review of Methods Used

In my study, methods typically applied in the study of plain pottery were adapted to tiles, namely, the study of forms, augmented in a small part with the study of fabrics or composition. This approach emphasises looking at tiles as objects comparable to pottery, where types and forms have for a long time been the primary focus of study. Within this category, plain or coarse ware pottery, as it is termed in research, would be the closest comparison group in terms of production processes, and therefore also methodology. As with roof tiles, they are objects for everyday use, usually locally to semi-locally produced, semi-domestic clay-based fired products requiring moderate but not excessive manufacturing skills and are common finds in excavations and surveys. In a comparable situation to roof tiles, coarse ware pottery was a neglected category of materials which has recently seen an expansion of interest as a field of study, with significant results focusing on morphology, transmission of ideas and technology, and transmission of materials and trade (Fulford, 2005, Joyner, 2005, p. 547).

Although acknowledging the importance of the analysis of technical characteristics of the raw materials used in production, the main focus of my research was on the forms and types rather than on provenance and technology. This in part led to typology becoming the main interpretive framework applied in the study. Another factor was a more practical one, that of the lack of a suitable technical infrastructure available to support more advanced scientific studies on e.g. composition or mineralogy outside a very basic macroscopic study of the fragments necessary for all clay-based finds (Whitbread, 2017). However, one experiment was made with the Paliambela material in order to test the applicability of this method for clay roof tiles. In this case, X-ray fluorescence (XRF) was used to analyse the chemical composition of the tile fabrics (Hamari et al., 2019). The methodology and the results of that study are explained in more detail below.

In addition to treating tiles as individual objects in need of sufficient documentation, this study regards it as important to study complete assemblages of tile finds (and other ceramic building materials), especially in cases where the assemblages derive from excavations. Conventionally, the term “assemblage” is applied to a collection of artefacts or ecofacts recovered from a specific archaeological context – a site, an area within a site, a stratified deposit, or a specific feature such as a ditch, tomb, or house (Joyce & Pollard, 2010), and it is this conventional meaning that is used in this work. In practice, this means a cross-cutting look at the tile material in its entirety, that does not overlook plain pieces or dismiss volumetric information. In particular, as in pottery, usually the whole needs to be looked at to understand the value of the assemblage and the individual items it consists of (Ikäheimo, 2003). In the object-oriented approach, which adopts its methodology from pottery studies, there is value in types as well as in quantities, volumes, and relative frequencies, which can only be traced when the whole assemblage is considered. This has direct implications for the development of documentation and retention practices.

6.2 Compositional Analysis

Although the key methodological approach in my research was related to tiles as objects and their morphology, there are methods related to the technical analysis of the tiles that are also relevant and even necessary in the study of ceramic building materials. It is increasingly clear that details related to clay fabrics and petrology are indispensable in all research related to fired clay objects – in pottery studies, they are a necessary element in understanding typology and groups (Buxeda i Garrigós & Madrid i Fernández, 2017; Roux, 2019; Whitbread, 1995) and can be used in determining provenance, production, and trade (Devolder & Lorenzon, 2019; Mills, 2013).

The justification of suggesting that such an approach would be possible in ceramic roof tiles is found in their nature and quality as earth-based materials. Roof tiles belong to the category of better made simple clay objects, as they need durability and non-permeability to be waterproof. This requires more advanced clay processing methods, such as levigation, clay mixing, and the addition of temper, as well as the application of slip, which consequently renders tiles more suitable for analyses. Tiles and bricks are in this sense closely related to pottery, with which they share a number of technological and material properties (Fragnoli, Boccalona, & Liberotti, 2023; Quinn, 2013, p. 213).

Research applying compositional analysis to the study of ancient ceramic roof tiles has so far been limited; only a scattering of studies in this field has previously been published, performed with a range of scientific methods from widely dispersed locations in the Roman empire (see e.g. Weaver, Meyers, Mertzman, Sternberg, & Didaleusky, 2013, with references). Despite being limited, these studies suggest that applying such methodologies to ceramic roof tiles is as viable as it is with pottery or other intentionally processed clay materials. A recent review article points out the range of different techniques available and their restrictions. It also emphasises the potential of material analysis in shedding light on the whole life cycle of earth-based building materials, from the procurement and manipulation of raw materials for the end products up to their use, discarding, and degradation (Fragnoli et al., 2023).

For the full set of assemblages and fragments included in this research, it was not possible to do a more detailed technological analysis than describing the fabrics in a standard macroscopic way, included as fabric descriptions in the catalogues. Rapid evaluation of fabric properties in the field was achieved by studying hand specimens using a magnifying glass at the time of the study; currently the preferred hand specimen documentation method would be a handheld digital USB microscope with sufficient magnification powers, capable of scale photographs preferably from fresh breaks. In most primary assemblages, the description followed the framework provided by Sanders (1999, pp. 477–478), covering hardness, colour, and texture (size, amount, and shape of inclusions). No proper identification of minerals was possible for any of the assemblages.

However, in the case of Paliambela, testing the applicability of a more advanced scientific analysis was possible, and is published with full technical details in Hamari et al. (2019). Of the two main branches of pottery analyses, petrography and chemical composition, chemical composition analysis was used. More precisely, a micro-portable XRF (pXRF) analysis was chosen because it promised to provide answers to a specific research question, that of the interrelation of the fingerline signatures on the tiles in the Paliambela assemblage (Hamari et al., 2019).

The goal of the analysis was to establish the composition of the fabrics and samples based on the results, and to compare these results or provenance groupings to the signatures, in order to understand how many potential workshops/places of origin could be distinguished, with the hypothesis that one signature would represent one workshop. In addition, a number of samples from other ceramic categories from the site, such as a different category of roof tiles (Early Byzantine) and relief-stamped bricks (Hamari, 2017a), were analysed to further clarify the internal categorisation of the material, and to provide a control group for the roof tiles with signatures. No comparative studies in provenance were possible as no suitable material or studies are available from the region yet. Compositional studies on tiles in general are rare, and none have been published on Roman period tiles (for some examples see Kilikoglou, Vassilaki-Grimani, Maniatis, & Grimanis, 1988 for Hellenistic period).

Based on the research questions outlined above, micro-XRF (μ-XRF) analysis was performed on a set of samples from tiles and bricks to determine their chemical composition. This method was chosen since it is non-destructive and provides a cost-effective way of obtaining preliminary results as first-phase testing of the hypotheses. The material analysed consisted of 22 samples: 16 samples of pan roof tiles and 6 of bricks. The samples were chosen from the catalogued excavation finds on the basis of their relation to a specific signature category (samples 1–12) or to a specific ceramic category (samples 13–22).

Although the method chosen was non-destructive, it was deemed suitable in light of the nature of the material and the needs of the analysis to extract specific samples from the material for analysis. The samples varied in size between ca. 1 and 3 cm³. They were cut from the tiles using a small metal wire handsaw. The external layer (a few mm) of each sample was removed by means of a drill and a tungsten carbide cutter, in order to eliminate possible surface contamination effects and to assure that the resulting clay surface was flat and smooth, with as little roughness and few anomalies as possible. A portable μ-XRF spectrometer was employed for the μ-XRF measurements. The ceramic standard reference material SARM69 was used for the quantification of the detected elements. Independent measurements at three different points on the clay were taken for each sample. The reported concentrations for each element represent the mean value of the corresponding concentrations measured at these three points (Hamari et al., 2019).

In order to compare and classify the samples, multivariate statistical analysis was conducted. The primary methods used were principal component analysis (PCA) and cluster analysis. Figure 7 illustrates the PCA plots for the measured samples for two different cases regarding the elements which are taken into account. In Figure 7a, PCA was conducted incorporating only the main elements, while in Figure 7b PCA was conducted with all detected elements taken into consideration (main and trace elements). From Figure 7, it is evident that the studied samples can be classified into four groups regarding their chemical composition, regardless of the elements taken into account during the analysis.

Figure 7

PCA of samples taking into account the elements: (a) Ca, Ti, Cr, Mn, Fe, Ni, Zn, K, Cu, and Si and (b) Ca, Ti, Cr, Mn, Fe, Ni, Zn, K, Cu, Si, V, Co, Pb, Sr, and Ga.

The results show that the majority of the Late Roman Laconian-style roof tiles from Paliambela belong to one compositional group. This also holds true for the macroscopically identified fabrics I and II, which have a similar chemical composition, indicating a common provenance despite the differing colour of the fabrics. The recognised Group 1 (Early Byzantine roof tiles) corresponds with the archaeological observation that this tile type is distinct and unrelated to the major tile type (Late Roman Laconian-style roof tiles), and most probably has a distinctly differing provenance from the rest of the material. Finally, the typologically unique trapezoidal brick also reveals a provenance that lies outside the major groups, possibly suggesting that it is an import.

The analysis shows that nearly all the signature types belong to one compositional group, and therefore have a common provenance. This is a significant result for the interpretation of tile signatures. The results suggest that signatures in this material were at least not used to denote the production of individual workshops. The analysis does not directly indicate whether this means that the tiles were produced at one individual workshop, or whether the material was produced in a geographically limited area where a number of production units were utilising similar raw material sources (see Whitbread, 1995 for discussion). The use of the same clay source by several production units cannot be ruled out, although usually the chemical composition is specific to one workshop, due not only to the individualised sources of raw materials but also to the specific methods employed for processing, mixing, and handling the clay raw materials (Tite, 2008). Therefore, the working hypothesis in that research paper was that the tiles with common provenance are most likely the product of one workshop or production unit. This suggests foremost that signatures did not equal individual workshops, as is commonly assumed, and a more nuanced view of their meaning needs to be adopted, possibly related to internal workflow and production control in the workshop(s).

The results also suggest that a more detailed study of the frequent and widely used signature types and their metrologies could produce very meaningful typological data on roof tiles. The results are significant enough to indicate the potential of compositional analysis in the study of roof tile signatures. p-XRF is methodologically a suitable approach, optimally supported by petrological and mineralogical analyses, and the increased development and portability of the equipment will in the future facilitate even further analysis. As always, more data and extended scientific methods should be employed to answer questions related to production and provenance more definitely.

7 Old Material, New Results

Section 6 outlined the key methodologies that proved to be useful in the study of plain ceramic roof tiles of the Roman period. Here I will present a summary of some of the key results reached by leveraging these methodologies.

Our understanding of the tile types in the eastern part of the Mediterranean during the Roman period rests on the established recognition of the Archaic tile systems. This study recognised that though not erroneous in its essentials, the dominance of this view has created a situation where tile finds are categorised in the general groups of Laconian (curving) and Corinthian/Hybrid (flat) pan tiles, and assigned only broad dating brackets (Hellenistic, Roman, Late Roman; Hellmann, 2002, pp. 298–299; Koskinas, 2011). Such a situation cannot be considered sufficiently nuanced in light of the results of the research, and when we look at these nuances, new results start emerging. A key factor behind the proposed tile areas is that the older systems stay recognisable and continue in use until Late Antiquity, and that they seem to have a geographically definable distribution areas that do not significantly overlap.

The material included in this research is not nearly extensive enough to function as a solid basis for a typology or typologies of plain tiles of the Roman Empire, but its cross-regional nature does open up possibilities in pointing our way towards the regional typologies of plain tiles in the East. It also points towards the elements that would be necessary to record in future field documentation as a means towards better tile typologies. Drawing together information both from the studied assemblages and published examples from the area, typological macro regions of tile traditions in the Roman period in the eastern part of the Roman Empire can be presented (Figure 8).

Figure 8

A schematic map of roof tile regions in the eastern part of the Roman Empire in the (Late) Roman period. Areas: A = Tegula tradition, B = Laconian-style, C = Roman Asia Minor, D = Cypriot, E = Cilician, and F = Near East general. Numbered points represent key and/or outlying findspots of representative tile types. 1 = Paliambela, 2 = Thessaloniki, 3 = Nikopolis in Epirus, 4 = Novae, 5 = Sobari, 6 = Labraunda, 7 = Sagalassos, 8 = Amorium, 9 = Kourion, 10 = Berytos/Beirut, 11 = Gadara, 12 = Gerasa, 13 = Petra. Base map: AWMC, CC BY-NC 3.0.

In addition to area A, representing the area of Western Tegula use, three macro regions (B, C, and F) in the Eastern Mediterranean consistently using a specific roof tile type can be distinguished in the Late Roman period. The first of the three macro regions identified encompasses the areas of Greece and the Southern Balkans, identified as area B in the map. In this macro-region, the Laconian-style curved pan and cover tiles dominate in the Roman-period find material, as exemplified by the tiles of Paliambela. Based on published finds, some smaller areas of distinctly identifiable regional types can also be identified here. These are the environs of Thessaloniki and Nikopolis, respectively (Hamari, 2019, pp. 93–94). The second macro region covers Asia Minor where the hybrid tile system was dominant, but with some Laconian- and Corinthian-style assemblages present, based principally on information deriving from published examples (Hamari, 2019, pp. 95–96). This area is marked as C in the map. The third macro region covers the Roman Near East, again with hybrid tile types but in more varied versions and with the absence of the Laconian-style types. This region is marked as F in the map and exemplified by the tiles from Petra. In addition, both Cilicia and Cyprus, marked as regions E and D on the map, formed regional centres with their own, identifiable types, distributing tiles widely, including down the Levantine coast (Hamari, 2019, pp. 96–97). It is emphasised that all these macro regions contain areas of significant regional variation, which is only partially recognised even in this research, let alone in earlier research literature.

In terms of genealogy or the development arc of the types, my research affirms that the types of tiles used in the Roman period were the descendants of the tile systems that came into use in the seventh and sixth centuries BCE in Archaic Greece and formed the basis of the tile types used later across the whole eastern Mediterranean. In particular, the research emphasised the adherence to the use of the specific combinations of pan and cover tiles – flat, curved, and gabled or semi-circular – that originally defined the Archaic tile systems called the Laconian, the Corinthian, and the Hybrid systems, continued to be followed throughout the period under observation. It is also important to note that in the Eastern Mediterranean, these types continued to be used regardless of the advance of the Roman Empire until the Late Antiquity (Hamari, 2019, p. 114).

In addition to typology, the chemical analysis done on the Paliambela tiles showed that, first of all, this kind of methodology can be successfully applied to plain roof tiles, and second, that the analysis provided interesting results to questions related to fingerline signatures. Chemical analysis performed with pXRF confirmed that most of these signatures belong to the same provenance group, meaning that they were probably manufactured in one limited area. This confirms that the different signatures do not directly represent different workshops, as has been proposed by earlier research, but rather provide information on the internal processes of a workshop (Hamari et al., 2019). This is considered particularly interesting in relation to the nature of ceramic roof tiles as products of small-scale crafts industries, where the choices of the craftsmen were presumably of primary importance in the stylistic and technical development of the tiles.

8 Conclusion

The previous analysis focusing on the methodology of the study of ceramic roof tiles clearly demonstrates the value of looking more closely at plain tile assemblages and individual tiles. Most, if not all, of the above outlined results are reliant on the sufficient documentation of the assemblages in excavations and surveys. This clearly implies that there is an urgent need to develop and mainstream the practice of documentation for tiles, especially in excavations, in order to create a large enough body of evidence to refine the emerging, more detailed typologies of the tiles of the eastern Roman provinces. This is a necessary effort, as local variations and specific contexts are our key to understanding the meaning and significance of the details that we see in any assemblage.

The minimum information necessary for ceramic building materials would be to document whether they are present among the finds at any given site. This kind of basic information can currently be found in many field reports. This study underlines that for any larger research questions, and especially for the purposes of developing a typology, this is not a sufficient level of documentation. An analogy with pottery has already been made throughout the text; in this regard the analogy would be that a field report would merely mention that pottery was “present” at the site, which would be an insufficient way of reporting in almost any context. Consequently, similar research strategies as those followed for pottery can and should be applied to tile and brick material.

A short attempt at defining a functioning documentation standard for roof tiles was created as part of my research (Hamari, 2019, pp. 137–145; see also Fragnoli et al., 2023 for guidelines on compositional and technical analysis). It is hoped that these guidelines make the documentation of roof tiles more frequent and more uniform across different projects. Ultimately, tile material can offer the same kind of interpretative potential as pottery does, for questions related to trade, economy, volumes of production, technology transfer, crafts processes, and cultural affiliation and agency. Ceramic tile materials should therefore be part of the standard materials research programme of archaeological projects. Additionally, results from individual research projects will help build up regionally meaningful pools of data, even if no direct results for a single site are expected. In all cases, the retention and documentation strategy for tiles should be planned beforehand alongside all other material categories, and then be actively implemented during the research. A very promising research initiative of the German Archaeological Institute has indeed started to look at the possibilities of digital documentation for roof tiles (Fless & Rheeder, 2021).

Approaching a fairly simple material group such as plain roof tiles from this angle serves as a reminder not to take for granted assumptions about the apparent homogeneity. The results demonstrate the considerable information value of roof tiles as source material. The ubiquity of the material makes it a suitable source of data for the analysis of economic processes such as production and distribution, and one that is able to shed light on the aspirations of both communities and individuals. Among all the valuable materials from the Roman world, plain earthen architectural elements have a lot to offer.