5.4. The Pottery by Steven Willis

5.4.1 Introduction

The survey resulted in the collection of a sizeable assemblage of pottery fragments which constitute the core dataset arising from the fieldwork. This assemblage amounts to 9871 sherds, weighing 46.023kg, all of which have been processed and entered into the database. As noted in 4.2, manifestly modern pottery was not gathered during the fieldwalking and hence the analysed assemblage constitutes material of prehistoric, Roman and medieval date, together with an element of more modern pottery identified during the processing of the sherds. The collected pottery assemblage was categorised following standard procedures, which are outlined here, with sherds being examined and assigned to exact fabric types. A total of 65 fabric types were identified. As would be predicted there was variation in the degree to which sherds were abraded both from field to field and between fabrics (dependent upon fabric hardness, etc.). Generally, however, the sherds were in good condition with negligible weathering. The average sherd weight per fabric type is listed in Table 5.9.

The following sections 5.4.2-7 consider the background of existing pottery study in the region and the nature of the material, methodology, chronology, fabric and manufacture, and forms and function.

5.4.2 Background: previous ceramic research and the nature of the pottery

The selection of the Ave valley region as the survey area was determined by the desire to compare cultural processes during the later prehistoric and Roman eras in a non-Mediterranean region of Iberia with those previously identified in the hinterland of Tarragona (Carreté et al. 1995). The Tarragona area is known to have been closely articulated with economic and social developments in the Mediterranean world during these periods, and to have had a strong tradition of pottery use, with archaeologically recovered ceramics being comparatively diagnostic and well dated. In contrast, the Ave valley region of north-west Iberia has received comparatively little archaeological attention away from its castros and the major Roman centre at Braga, and this was a key raison d'être in deciding to undertake the work in this locality. The lack of previous fieldwalking and limited modern study, however, meant that there was little existing knowledge of ceramic traditions and fabrics upon which to draw and base a typology of the gathered ceramics. To our knowledge, for instance, no kiln sites have been excavated in the region; hence the sources of much of the pottery of the region are uncertain. Similarly, archaeological investigations, other than at the castros and major Roman foci were, until recently, rare and so there is an absence of published stratified sequences or knowledge of the pottery of ancient rural settlements. These factors, in particular, impact upon our ability to date the Roman and early medieval pottery and to place it into a refined sequence. Consequently the project needed to establish its own typological recording system.

The typology and understanding of the chronology of the material presented here should thus be seen as, essentially, a provisional one, which future work in the region can develop and enhance. The cultural associations and dates of the 65 survey fabrics are listed in Table 5.1.

There are common difficulties associated with surface collected pottery that are endemic (e.g. small sherd size, abrasion, comparative rarity of sherds diagnostic of rim form, the unstratified nature of the material and the lack of association with other cultural and environmental evidence cf. Haselgrove et al. 1985), and from which the present collection was not immune. In addition, further characteristics of this material hinder confident cultural attribution in the case of a range of fabrics. First, there was an overall absence of familiar Roman and early medieval traded wares. Secondly, some regional fabrics, including the numerically most common fabric, 308, seem likely to have been current over a long time span.

More positively, the prehistoric pottery was readily identifiable occurring in a series of mica- and quartz-tempered wares with distinctive forms. Most of the latter material is Iron Age and is typical of the regional pottery of this era. The ceramics of this period have, exceptionally, been comprehensively studied and published as a result of excavations on castro sites (e.g. Almeida 1974; Martins 1988) and in synthetic studies (da Silva 1986; Little 1985; 1990). A range of well characterised fabrics associated with Roman form types were also discernible, together with a few sherds of familiar Roman traded wares, including amphora and sigillata. Dolia too, of Roman date, were present in distinctive fabrics (102 and 108). The absence of both coarse and fine-ware types common within stratified levels at Braga, Bracara Augusta (cf. Alarcão 1966; Martins and Delgado 1990b), was a striking pattern (see 5.4.8). A small amount of material was diagnostically early medieval or medieval. Parallels for the fabrics (and in some cases forms) of this survey are noted under the individual fabric descriptions.

To summarise, our understanding of the typology and chronology of the survey pottery has been constrained by the nature of the material itself, in addition to the lack of existing studies of the ceramic traditions and consumption patterns of the region. These factors place limits upon the scope for analysis. A range of fabrics and types present are well characterised and their cultural associations and broad dates are known. Others, however are less diagnostic and appear to have endured over a long time-span. Analysis of the data, has, however, assisted in identifying the likely cultural association of a number of the less diagnostic fabrics.

5.4.3 Methodology

The survey pottery was processed in batches between 1996 and 1998 by means of the following procedure. The washed and dried sherds were sorted into exact fabric types following the general principles established by Peacock (1977), with refinements. The conventions and standards employed adhere to those outlined by the Prehistoric Ceramics Research Group (PCRG 1995). Fabrics were isolated and sherds categorised via examination using a x20 binocular microscope, supplemented by a hand lens and macroscopic identification. The different fabrics present were allocated unique numbers (e.g. Fabric 101, Fabric 305), with these numbers following a simple system of division. Blocks of three-figure numbers were ascribed to broad pottery classes, with the leading number relating to a principal distinctive characteristic:

The fabric categories are listed below by summary description; full descriptions are given in the Fabric Descriptions Appendix.

5.4.4 The fabrics

101-110 Micaceous wares
101 Highly micaceous ware, usually dark grey or brown, associated with hand- made forms
102 Very coarse oxidised ware, associated with large storage type vessels, especially dolia
103 Generally unoxidised ware with plentiful inclusions
104 Oxidised micaceous fabric, similar to 101 and associated with hand-made forms
105 Oxidised, fairly fine, fabric, associated with hand-made forms
106 Finer ware, generally un- or partially oxidised, associated with hand- made forms
107 Coarse, partially oxidised, heavily tempered ware
108 A very coarse oxidised ware, associated with large storage type vessels, especially dolia
109 Light grey micaceous fabric
110 Comparatively fine micaceous grey ware

201-216 Grey/Unoxidised wares
201 Very fine grey ware/s
202 Moderately coarse grey ware
203 Partially oxidised ware with consistently grey exterior surfaces
204 Distinctive quartz-tempered grey ware associated with thin-walled forms/small bowls
205 Grey fabric, probably the unoxidised equivalent to fabric 308
206 Light grey fabric, with comparatively few inclusions
207 Coarse grey ware with quartz temper
208 Not in use
209 Grey ware with quartz and mica
210 Very fine grey ware with burnished surfaces
211 Fairly fine grey ware
212 Distinctive grey ware with few inclusions
213 Fine grey ware fabric
214 Coarse ware with pale grey surfaces and pale red core
215 Grey ware with red core
216 Partially oxidised coarse ware

301-330 Oxidised wares
301 Very fine light red ware
302 Generally red, high fired fabric group, mainly, if not entirely, modern
303 Buff fabric with occasional quartz fragments
304 Red coarse fabric
305 Late Roman African Red-Slipped ware/Clara D/TS H or equivalent
306 Light red fine fabric with burnished surfaces
307 Coarse light red fabric
308 Light red fabric, associated with wheel-made vessels
309 Mid and dark red fabric
310 Hard fabric with yellow brown-brown surfaces
311 Light red fabric with few inclusions
312 Light red fine fabric with thin glaze
313 Pink fabric with reddish-yellow colour-coat
314 Buff/yellow fabric with slip
315 Red, hard fabric with slip
316 Buff to light red fabric with few inclusions
317 Fine buff fabric
318 Generally red coarse fabric, similar to fabric 307
319 Red, comparatively fine, fabric
320 Pink fabric, comparatively fine
321 Buff-pink colour-coated ware
322 Fine red fabric
323 Coarse, somewhat micaceous fabric
324 Oxidised coarse ware
325 Fine oxidised fabric, associated with thin-walled vessels
326 Distinctive fine red fabric
327 Red fine ware fabric, similar to fabric 305
328 Central Gaulish Terra Sigillata
329 Pale pink fabric
330 Coarse, distinctive, somewhat micaceous fabric

401-404 Heavily quartz-tempered fabrics, associated with large vessels
401 Very coarse, partially oxidised fabric
402 Very coarse, partially oxidised fabric
403 Oxidised coarse fabric
404 Reddish brown fabric

501 All glazed wares, bar fabric 312

601-604 Coarse wares with frequent tempering with granite and quartz
601 Oxidised fabric with very coarse inclusions
602 Variably oxidised fabric with quartz and mica
603 Variably oxidised fabric with quartz and biotite mica
604 Moderately coarse buff fabric with quartz/granite inclusions

701 Amphorae
701 Baetican amphora fabric

The processing and recording of the pottery proceeded on a field by field basis (Query pottery quantities by field). Once the sherds from a specific field had been divided into fabric types they were recorded using a standardised proforma. The documented information is reproduced here (Query pottery details). Within each fabric, featured sherds (i.e. from rims, handles, bases, items with decoration, and so forth) were itemised individually, while body-sherds were often grouped together. The sherd type was recorded (i.e. whether a sherd was from the rim, base, handle or body of a vessel) as was the vessel form, where identifiable. In practice, identifying the vessel form from which a sherd derived, even a rim sherd, did not prove straightforward. This was partly because of the small size of the sherds, which often represented only parts of rims and/or small percentages of the rim circumference, thus making it difficult to ascertain vessel mouth diameters with any confidence. Additional problems arose from the often elaborate or seemingly idiosyncratic forms present (see examples). All of the sherds were counted and weighed and, where possible, rim and base diameters were also measured to establish the Estimated Vessel Equivalent (EVE). For this purpose, the percentage of extant rim or base was recorded as the Rim Equivalent (RE) and Base Equivalent (BE) figure respectively, where an RE of 1.00 would represent a complete circumference. The occurrence of surface slips or glazes was recorded, as was the incidence of decoration (Table 5.10). Information was also collected on surface finish, specifically where burnishing or significant smoothing was discernible. Traces of use, in the form of surface residues, were also noted, for instance, where soot or carbonised deposits occurred. Other points of significance were recorded under the 'Comments' section. Items to be drawn were then selected; drawn sherds are accordingly noted in the database.

In consequence, the database contains quantitative information derived from three different methods, namely count, weight and Estimated Vessel Equivalent (EVE). This increases to four if the number of rims present by field and fabric, which calculation is possible from the dataset, is included. Table 5.9 summarises the quantities of each pottery fabric recovered in the course of the survey. Table 5.10 summarises the incidence of decoration by pottery fabric and Table 5.11 presents information about the quantities of the different ancient tile fabrics that were collected.

5.4.5 Chronology

Table 5.1 documents our assessment of the date of the fabrics. The fabrics are categorised in terms of their cultural association, with these categories covering broad time spans. Where there is doubt about this, fabrics have been listed in Table 5.1 as having 'Possible' date ranges. As noted, there was no existing regional pottery typology covering the extensive range of types recovered during the fieldwalking; hence the survey typology and our chronological understanding of the fabrics represent, in some respects, an initial study. Several of the constraints to establishing the chronology of the collected pottery were outlined in 5.4.2.

A fundamentally important factor for the survey, and, indeed, for other archaeological considerations of the region, are the marked cultural differences between north-western Iberia (including the Ave valley and its hinterland) and other regions of the Peninsula, which endured through later prehistory and the Roman era (cf. Keay 1988; Millett forthcoming). Amongst other manifestations these cultural differences are apparent in the varied traditions of production, importation and consumption. Equally, the extent of archaeological knowledge of the sequences varies between regions. In general, Roman and later pottery from other areas of Iberia is better characterised than in the north-west. The survey team brought to the project a strong understanding of the ceramic history of various parts of Iberia (particularly that of southern and central Portugal, and north-eastern and southern Spain), while one member of the team (Francisco Queiroga) had extensive knowledge of the prehistoric pottery of northern Portugal. The particular area in which we needed to increase our understanding was that of the local ceramics of the Roman and early medieval periods.

Understanding of the chronology of the fabrics and types encountered amongst the survey material draws upon various sources of information. These sources include the typology of the material and its technology of production, reference to published material, museum collections, recently excavated stratified groups and, importantly, discussions with Portuguese colleagues, as well as observations upon the occurrence of material, and particularly of associations. In essence, these are standard means by which an awareness of the ceramic sequence within a region can begin to be constructed. The existing literature and knowledge are certainly strong for specific periods and localities. However, a more refined chronological understanding of a number of the fabric categories is desirable, and doubtless this will come with future work and publication in the region. As previously noted, analysis of the database has assisted in identifying the likely cultural association of a number of the less diagnostic fabrics (see section 5.2).

Examination of pottery collections in the regional museums was undertaken as the survey material was being processed and this proved extremely helpful. Museums visited included those at Barcelos, Braga, Esposende, Guimarães, and at the Citânia de Sanfins and the Unidade de Arqueologia da Universidade do Minho. A series of sites were visited where excavations were underway (e.g., in Braga, and at the castros at Penices and Terroso). In addition. a number of Portuguese archaeologists made recently excavated stratified groups available for study. Assemblages from Professor Queiroga's fieldwork at a range of sites in the region was also to hand. The debt of gratitude owed to a number of archaeologists and pottery specialists based in the region who kindly gave advice on the identification and dating of material is acknowledged. Some chronological information relating to specific fabrics is included with the fabric descriptions.

A number of coarse oxidised or variably oxidised fabrics containing frequent granite and quartz tempering (Fabrics 601-604) may represent the earliest pottery identified amongst the survey material. Sherds belonging to these categories were infrequent finds but on typological grounds are evidently prehistoric. Fabrics with such tempering in the region are known from Bronze Age sites (e.g. Little 1990) while the pottery from Chalcolithic period sites in the region is likewise qualitatively similar.

One potentially useful chronological index independent of the pottery is the incidence of tegulae amongst the survey material. Fragments of tegula were regularly found to occur amongst concentrations of collected pottery sherds (though the meaning of their presence amongst surface groups with regard to dating should be assessed with care). In north-western Iberia, Roman- style tegulae occur at castros and at rural sites from c. AD 50. The type continued in use beyond the Roman era but appears not to have endured beyond the end of the 12th century.

The publication of excavations undertaken in recent years, especially rescue excavations in the countryside (e.g. in advance of the new road developments) as well as at the long-lived urban centres (such as Guimarães and Vizela) and at the smaller settlements, will begin to fill in gaps in the regional ceramic chronology, via the presentation of stratified associated groups of pottery. However, given the nature and longevity of pottery traditions of the region, and the infrequency of imported wares (as highlighted by the present survey) it is uncertain whether the refined chronological understanding that is now possible on the basis of survey collections for many locations around the Mediterranean littoral will be achievable for this part of northern Portugal. This point is perhaps one of the most significant findings of the survey. It should not be regarded as grounds for pessimism, though, as there is unequivocally so much of archaeological value to be learned from the Ave region, which the chronological information to hand will readily facilitate.

5.4.6 Fabric and manufacture

Independent of fabric and cultural association, sherds recovered during the survey indicated a high level of skill, care and control in pottery production. Local fabrics are dominant. The prolific fabric 308 is not sourced but seems very likely to be of regional origin, given that similar clays have evidently been used in pottery making in the 20th century in the region (cf. collections at Barcelos Museum). Manufacture of this ware is likely to have occurred at more than one location in the area. Several other fabrics amongst the survey assemblage seemingly constitute variations of 308: 301 is a fine ware version associated with thin-walled vessels, while 307 is a somewhat coarser version, and 205 and 202 may be unoxidised examples. Hence the wide range of fabrics represented amongst the collection may conceal the fact that many have shared sources. Fabric 101 and others in the 101-110 and 601-604 ranges seem certain to be local or regional in origin. Fabric 101 contains muscovite mica with quartz, presumably derived from the mica schist. Clay sources occur in the region's valleys.

Table 5.1 documents the technology of production associated with each fabric type represented amongst the survey material. Most of the sherds in the fabric ranges 101-110 and 601-604 (with the exception of the dolia) seem likely to be hand-made, though they are more or less invariably well formed and finished. All, or virtually all, of the other fabrics represented amongst the survey assemblage are associated with wheel-made forms.

5.4.7 Forms and functions

One of the over-riding patterns amongst the survey pottery is a very strong form-fabric correspondence, in so far as specific fabrics are associated with particular forms and ranges of forms. This phenomenon is borne out by the illustrated sherds. A wide range of forms is represented amongst the material. The (inevitably) highly fragmentary condition of the pottery, though, means that a high proportion of the sherds are too small to ascertain their forms, profiles and sizes with confidence.

Of the fabrics associated with pre-Roman pottery types only fabric 101 is sufficiently well represented to permit any assessment of the common forms and possible functions. This fabric type is evidently well represented at the castros of the region and is clearly akin to material examined by Little (1990). Amongst the assemblage from the project, wide-mouthed forms are frequent, and evidently comprise bowls; in some cases these are very large indeed. Perforated sherds in this fabric may well come from hanging bowls. A lid in this fabric was recovered from Terroso. Both thick- and thin-walled vessels were manufactured in this fabric; in other words it was employed to produce both fine wares and robust types. The trends seen in the material from our survey are consistent with the wider identified patterns associated with the dark micaceous Iron Age wares of northern Portugal. The deliberate cultural preference for this highly micaceous fabric in the production of a wide range of forms may have been determining, over-riding the potential utility of using other inclusion types or 'recipes'.

Reference to the traces of vessel use data in the database verifies that, significantly, sherds in fabric 101 show little in the way of soot or carbonised residues upon their surfaces. This suggests that these vessels were not used over fires.

Amongst the fabrics associated with Roman era pottery (e.g. fabrics 303, 305 and 308), dishes and bowls are common. Fabric 305 is a highly standardised fine ware which occurs more or less exclusively in a range of dish and shallow bowl forms. The exact form of these items is difficult to parallel rim by rim but they seem certain to be regional manifestations of the red-slipped fine ware tradition of the metropolitan Roman world; the forms are similar to Beltrán Lloris's form 617 (Beltrán Lloris 1978). In contrast to some of the fabrics associated with Iron Age and prehistoric types, the Roman and later pottery of the region is almost invariably oxidised.

There is a high frequency of carbonised deposits or soot on the surfaces of sherds in fabrics 301 and 201. This appears surprising, given that these fabrics were employed, more or less exclusively, to manufacture thin-walled fine vessels.

5.4.8 Discussion

The infrequency of Roman traded wares was a striking feature of the collected material, though their presence in such small quantity is a feature of north-west Iberia generally (cf. Naveiro López and Pérez Losada 1992). Amphorae were virtually absent from the collected assemblage. Large quantities of amphora sherds were not anticipated from the area at the inception of the project, given the lack of reports of such material within the existing literature. Nonetheless, the paucity of finds belonging to this category demonstrates unequivocally the virtual absence of these transport and storage vessels from the areas sampled. They were evidently neither produced nor traded in these rural locations in any number, nor were their contents consumed. Even at Braga, Roman amphorae have been far from prolific site finds; the total excavated assemblage from the city is small, early in emphasis and dominated by South Spanish fabrics in Haltern 70 and related forms (Rui Morais, pers. comm.). Dolia, by contrast, were found to be comparatively widespread across the survey area, occurring in standardised types.

The absence from amongst the survey assemblage of many of the coarse and fine- ware fabric types common within stratified levels at Braga was noted above. Although a consistent trend in this case, this phenomenon of difference between the Roman pottery from an urban centre and that of its hinterland is paralleled elsewhere in the Empire, for instance between York and its surrounding settlements (Roskams 1999). Evidently urban and rural consumers often had differing patterns of ceramic consumption, doubtless due to contrasting cultural norms and variations in ceramic supply.

Pottery fragments were more frequent than tile fragments, though as the survey data show, groups of pottery and tile were often found in association.

Query pottery data from first walk | Query pottery data from second walk
Query tile dataset from first walk | Query tile dataset from second walk


© Internet Archaeology URL:
Last updated: Sat Dec 30 2000