The establishment of a national reference collection for pottery of the Roman period has been discussed since 1972, when it was proposed by Dame Kathleen Kenyon at the first meeting of the Study Group for Romano-British Pottery held in Oxford (Detsicas 1973). Informal and sporadic discussion of the need for a collection has continued since, particularly after the publication of the Guidelines for the processing and publication of Roman pottery from excavations (Young 1980). In March 1986 a meeting was held at the British Museum to discuss the feasibility of establishing collections for all periods. While there was agreement on the need, and that separate collections should be maintained for the different periods, no further action resulted.
The results presented here stem from the survey of the Current state of Romano-British pottery studies undertaken by Professor Mike Fulford at the request of English Heritage, which made explicit recommendations for the establishment of both national and regional fabric collections (Fulford & Huddlestone 1991, 52). In November 1992, at an EGM of the Study Group for Roman Pottery, it was agreed that a national collection should take precedence, providing a basis and model for regional ones. The initiative and full funding for this project (National Roman Fabric Reference Collection – NRFRC), including this publication, was provided by English Heritage. The work was undertaken by the Museum of London Archaeology Service with support from the British Museum (Departments of Prehistoric & Romano-British Antiquities and Scientific Research).
This project has recently been outlined in Britannia (Tomber & Dore 1996) in order to summarise and to draw public attention to the collection. In rare instances details given there may differ, and this handbook should be considered the definitive statement.
Research into Roman pottery has a long tradition, rooted in typological studies of the 19th century and exemplified by scholars such as Dressel (1891) and Dragendorff (1895; 1896). While a typological approach has served us well, other variables – including fabric – are interdependent and equally important. This has been recognised for some time and is apparent in early Romano-British pottery studies, for example in the work of John Gillam (1939; 1951; 1970), where macroscopic fabric differentiation was important in the subdivision of pottery types; it is also reflected in some descriptions in the Student’s guide (Webster 1976). Although scientific techniques were sporadically applied to the study of Roman pottery fabrics as early as 1943 (Lais 1943) and again in the work of Gillam (1963), this approach did not gain the immediate popularity as seen in other geographical areas or chronological periods (see Rice 1987, 310–13 for a review).
In the late 1960s and early 1970s fabric analysis became more prominent in Romano-British pottery studies, due to the work of Professor David Peacock. Through the application of geological thin-section techniques to pottery of the Roman period, Peacock had particular success in the sourcing of pottery types, initially on pottery from the Malvern region (Peacock 1967) and subsequently concentrating on amphorae (eg Peacock 1971). This work served as a model for additional scientific analysis, including important studies on, for example, black-burnished (Williams 1977) and Dales (Loughlin 1977) wares.
While the adoption of scientific methods provided a systematic basis for the study of pottery fabrics, equally important was the generalisation of these techniques to low power microscopy or macroscopic observation. Basic guidelines were drawn up and applied by Peacock, together with good common sense in the macroscopic sorting of large pottery groups (Peacock 1977a), and the value of this approach was demonstrated by the publication of a large volume of pottery from Carthage (Fulford & Peacock 1984).
In Britain the initial response to fabric analysis was somewhat divided, being either greeted with great scepticism or wholeheartedly embraced, as a panacea for all ills of Roman pottery. The past twenty-five years have seen numerous trends in the adoption of fabric analysis and it is hoped that we have now achieved a balanced approach, in keeping with Peacock’s original aims, which recognises the importance and limitations of fabric analysis within a broader framework (see also Young 1980, 4–5). Fabric should be seen as a single variable of ceramic studies, to be integrated with vessel form/function, archaeological context and overall research priorities (eg Shepard 1976). The true value of source characterisation and fabric differentiation through scientific techniques is that it provides a firm basis by which pottery can be placed within its wider context and therefore be used for more generalised interpretation (Renfrew 1977). This is admirably illustrated by several collections of multi-period papers which use fabric analysis to investigate wider ceramic issues (Freestone et al 1982; Middleton & Freestone 1991; Peacock 1977a; Williams 1983, for a review).
Nevertheless, limitations are inherent in both the type of fabrics that are amenable to analysis, and in the methods used. Many fabrics, particularly those originating in southern Britain, will be composed of common minerals (ie quartz, limestone, flint). These fabrics will be difficult to distinguish from each other and virtually impossible to assign to source, even with the aid of scientific techniques. Where complex geology – more amenable to source characterisation – is involved, macroscopic or low power microscopy will be inadequate for detailed identification. More precise techniques, such as thin section, heavy mineral and chemical analyses provide a necessary means for confirming and refining fabric groupings established by other methods, as well as providing information on technology and source. Petrological techniques provide an efficient and inexpensive first step in the scientific definition of fabrics, particularly for identification of rocks and minerals, and form a good basis for the selection of additional analysis where appropriate.
The importance of scientific techniques has been recognised by both the Department of Environment, and latterly English Heritage. Their commitment to scientific support for government funded excavations has been demonstrated since the mid 1970s by the establishment of a Ceramic Petrology Fellowship, a post held by Dr David Williams at the University of Southampton. More recently English Heritage has funded a series of two-day courses in ceramic petrology taught by David Williams at the University of Southampton.
Despite the recognition of the importance of fabric analysis, Fulford recorded that 30% of pottery reports were written without the aid of a fabric reference collection (Fulford & Huddlestone 1991, 20). Notable exceptions do, of course, exist. The first and largest fabric reference collection in Britain was that established by the Museum of London’s Department of Urban Archaeology in the mid 1970s as a research tool into the pottery of London from all periods (Orton 1979; Rhodes 1977). Organised by dominant inclusion, its success rested on their accurate identification. In contrast, the NRFRC provides a synthesis of current research and, while recording inclusion type, is organised by established fabric groupings.
The purpose of the NRFRC is to provide an infrastructure for future research into Romano-British pottery. More specifically, the NRFRC provides a standard for the identification and description of Roman pottery types, which will be of value to all those who work with Roman pottery. This is achieved through the physical reference collection as a means for fabric verification, together with consistent and standard descriptions of these sherds. The dissemination of standardised fabric descriptions in this handbook is intended to eliminate repetitive work and duplication in publication, while at the same time furthering consistency in identification through consistency in description. By establishing standard nomenclature, communication between pottery practitioners in different regions should be improved and comparative studies advanced.
In addition, this handbook provides a barometer on Roman pottery studies, with the current state of research reflected by the bibliographic data. The compilation of the NRFRC has drawn attention to those areas most amenable to and under-represented in scientific analysis, such as kiln products and mortaria. It will also help to isolate those areas which are most poorly understood, and towards which future research should be directed. With less time spent on the description of well-known wares, important regional issues of production, distribution and marketing can take priority.
As a permanent institution with curatorial support and free access, the British Museum was designated as the most appropriate location to house a national collection. While travel to London is expensive, it remains more accessible than elsewhere in the country. Other collections within the British Museum complement the NRFRC, including a medieval fabric collection in the Department of Medieval & Later Antiquities, which will also be the recipient of an Anglo-Saxon stamp corpus. During the planning stage of this project in Autumn 1992, Dr Ian Longworth, the then Keeper of Prehistoric & Romano-British Antiquities generously supported a request from the Study Group for Roman Pottery to house the project in his department and offered the full assistance of his staff. This support has been maintained by the present Keeper, Dr Timothy Potter, with great commitment and enthusiasm. All aspects of the project were greatly facilitated by Valery Rigby, who as both a member of staff of the British Museum and – when the project was initiated – the President of the Study Group for Roman Pottery, was instrumental in bringing the project to fruition. Finally, the Keeper, Dr Sheridan Bowman, and Dr Ian Freestone of the Department of Scientific Research undertook to house a library of thin sections within their department and to assist in the analysis of some of these samples.
There are three main components to the fabric collection:
The explicit intention was to incorporate both fabrics imported into Roman Britain and those that travelled throughout the province or between major geographical or ceramic regions; in both cases these were normally restricted to types with a known source. In this, and in all other respects, Vivien Swan’s Pottery kilns of Roman Britain (Swan 1984, particularly maps 13–18) formed an invaluable resource and should be consulted for additional information on individual kilns or fabrics.
Every effort was made to be consistent in the definition of fabrics to be included, although some difficulties did arise – both of a practical and more theoretical nature. The most obvious problem occurred for types too rare for the donation of sherds; more difficult were those types which could not be accurately defined or provenanced. Occasionally unprovenanced types with a widespread distribution (eg grog-tempered and shelly wares) were included, to illustrate different technologies and their commonly accepted terminology.
The definition of imported wares was fairly straightforward, but there were obvious problems in isolating wares that travelled between regions – as they may well travel but not be identified. On one level our selection policy will serve to reinforce this, but within the financial and time constraints it was not possible to be totally comprehensive in selection; mechanisms do exist, however, and are described below, to add types when they are better understood or become available. Simplification of some fabrics will have undoubtedly resulted, but certain complexities are better tackled on a local level, where the range of variability will be greater.
Equally important was the selection of samples, and every effort was made to ensure that the sherds included were properly identified, particularly when kiln samples were not used. Should kiln samples become readily available at a later date, these will be added or substituted for the current samples. Sample number was limited by both time and space, and therefore the full range of variability could not be represented: emphasis was instead placed on typical sherds. The NRFRC is intended purely as a fabric reference collection, and form was not a main criterion in sherd selection. Nevertheless, every attempt was made to provide a limited range of forms and decorative motifs, although again this was governed by availability of material.
At the inception of the project a questionnaire was sent to all museums with known collections of Roman pottery, based on a database provided by the Early Department of the Museum of London, and the results from this survey helped to identify potential donors, as well as providing information on current collections. A large number of samples was provided by the British Museum and donated by the Museum of London, but other individuals and organisations throughout the United Kingdom and abroad were also generous with material. Pottery specialists from the individual regions were instrumental in the collection and verification of samples, while particular wares were selected or approved by acknowledged specialists: mortaria by Kay F Hartley; samian by Brenda Dickinson and Joanna Bird; and Gallo-Belgic wares by Valery Rigby.
Fabrics were classified on two main levels. Firstly they were allocated to ware groups (distinct from, but sometimes utilising the same terminology as technology group, below), defined as samian (SAM), fine wares (FINE, incorporating ‘classic’ fine wares and other slipped sherds), coarse wares (COAR, both oxidised and reduced) and amphorae (AMPH). All fabrics were described on pro formas within these broad groupings. At a more detailed level, fabrics were examined and described within technology groups which were defined by a diverse range of factors, including surface treatment, and are listed below. Notwithstanding the obvious problems resulting from the abrasion of surfaces, the selection of surface treatment as a criterion was not without other inherent problems, as many industries produce a single clay matrix with colourcoated, mica-dusted or other surfaces. Nevertheless, it was considered a useful working device by which to describe and define fabrics.
For reference purposes each fabric has been given a Common Name, abbreviated to a short code. The full name sometimes includes qualifying information, shown in brackets, normally to aid recognition of subdivisions which are otherwise distinguished only as numbers. The coding system has been designed for speed and ease of use with a computer database and to provide a universally applicable scheme which can generate new codes in the future. The complete code may be up to eight characters long, divided into the following three fields, each of which can be searched separately:
Inevitably, not each grouping comprised an easily defined, discrete category. Particularly difficult were the distinctions between red-slipped and colour-coated wares, and the placement of vessels decorated with roughcasting. In the former case, industries producing primarily vessels with a red-brown slip or copying samian forms or Raetian-style mortaria were coded as red slipped, other slipped vessels as colour-coated wares. Roughcast vessels may be included with either colour-coated or oxidised wares. The following categories were defined:
Each fabric is described following the system proposed by Peacock (1977), including variables of colour, hardness, fracture, inclusions, sorting, quantity and rounding. Similar guidelines, containing useful illustrations, have been adopted for prehistoric pottery by the Prehistoric Ceramics Research Group (1995). Detailed pro formas (Fig 1) giving the characteristics of each sample, which was assigned a unique number, are housed in the archive. Other distinguishing features, relating to surface treatment, decoration and sherd type, are included where relevant. These variables (together with computerised keywords) are listed and defined in Appendix 1.
The limited and widely available range of inclusions, such as quartz and limestone, means that many Romano-British fabrics can prove difficult to distinguish from verbal descriptions alone. This is exacerbated by wheel technology, which limits the size of inclusions, making macroscopic identification difficult even when diagnostic fragments are present. Fabric descriptions are therefore based on both the hand specimen (using X20 binocular microscope) and in thin section under the petrological microscope. The former aims to describe and identify what can be seen under the binocular microscope, but where inclusions cannot be readily identified this is a purely descriptive exercise. Each fabric was therefore supplemented with a thin section from a typical sherd, allowing more precise identification of inclusions, and the notation of additional significant features only visible under the petrological microscope. The differing levels of resolution for the two techniques have meant that the description in the hand specimen and in thin section will not necessarily correspond exactly. In most cases only one sample was examined in thin section, excluding any discussion of variability.
The approach to the thin sections varied somewhat depending on the type of geology involved: where distinctive inclusions are present identification has concentrated on those diagnostic features, rather than itemising all material visible in thin section; elsewhere, when only common inclusions are present they have been described in full. Clay was not described, but a note was made when isotropic. Similarly, if quartz is not described it can be assumed to be monocrystalline, and is normally only specified for polycrystalline quartz. Additional thin sections of our fabrics (but not our samples) may exist elsewhere in the country, and this information will be contained in a database of all thin sections held in the UK currently being compiled by a team headed by Alan Vince and funded by English Heritage.
Although all production sites exhibit some degree of variability, descriptions published here concentrate on the ‘standard’ fabric rather than outliers. Much of how we identify fabric relies on intuitive recognition, and the recording of a standard set of variables is one method by which to make this intuition explicit. Written fabric descriptions can often be identical for fabrics that we can easily discriminate in the hand. In an effort to overcome long lists of indistinctive features, we have highlighted in bold those features most useful for identifying and distinguishing fabrics; colour photographs have also been used to supplement definition. In addition, some characteristics that occur nearly universally in our samples (for example being wheelmade) are only included when they deviate from the norm: this is explained in full under ceramic variables below.
For examination and recording, in the first instance, the collection was divided between the two authors, each independently recording a portion of the fabric samples. Following this several intensive sessions were held, where the authors jointly reexamined the majority (c 95%) of the samples, and critically reviewed the standard and consistency of recording, correcting and expanding where necessary.
Mr John Cooper of the Natural History Museum examined our shelly fabrics and identified inclusions in the hand specimen, while selected thin sections were examined by Sylvia Humphrey, Ian Freestone and Andrew Middleton of the (then) Department of Scientific Research, the British Museum
Variables are described below, with further definition given in Appendix 1 as appropriate.
Colour is defined by both Munsell colour values (Munsell 1992) and free descriptive terms (not those employed by Munsell). The consistent application of the Munsell charts has frequently been called into question. In this respect it is interesting to note that the two authors independently selected the same Munsell chips, while it was necessary to discuss free descriptive terms before reaching an agreement on their usage. Colour photographs of fresh breaks illustrating each fabric (but not necessarily each surface treatment represented in the same fabric) also provide another indication of the typical colour, although this method is not without its own problems. The most difficult tones to reproduce accurately were the reds, as is most obvious in the case of samian. Nevertheless, photographs give an additional visual dimension, used to create a general impression of the fabric, since an individual break is unlikely to include all the components described, particularly at the lower level of magnification (X2) considered appropriate for publication.
Hardness was recorded according to Peacock’s schema, adopted from Mohs’ hardness scale and comprising ‘Hard’, ‘Soft’ and ‘Very hard’.
Surface feel is described as ‘Rough’, ‘Harsh’, ‘Powdery’, ‘Smooth’ and ‘Soapy’ and may be affected by soil conditions.
Fracture in fresh break is noted as ‘Conchoidal’, ‘Hackly’, ‘Finely Irregular’, ‘Laminated’ and ‘Smooth’.
Inclusion composition. Wherever possible, inclusions have been identified in the hand specimen. However, as noted above, it can be misleading to attempt to define certain inclusions at this stage and in many cases descriptive terms are more appropriate. Included in this group are complex rock fragments, which can sometimes be assigned to a general class (ie metamorphic, sedimentary or igneous), but may frequently only be classified as ‘ROCK’. Thin-section examination serves to refine these descriptions.
Inclusions with a high iron oxide or clay content can also be problematic to identify in the hand specimen. Iron ore and iron-rich clay pellets are most likely to fall into this category. Although they can be distinguished in the hand specimen by the use of a magnet, they are frequently small (eg <0.1mm), hindering precise identification: therefore most inclusions of this nature have been recorded as iron-rich inclusions (IRRI). Nearly all fabrics contain a sprinkling of this matter and while they are rarely significant in distinguishing the fabric they are included both for the sake of completeness and to enable the description to ‘tally’ with the sherds. Clay-rich inclusions (pellets) and metamorphic rocks may also be difficult to distinguish from each other and have sometimes been recorded as argillaceous inclusions (ARG). Larger clay pellets, particularly those that are quartz-rich, can be more readily identified in the hand specimen, and are frequently duller in appearance than iron ore. Although not necessarily clear cut, identification is more precise in thin section and, apart from rock inclusions, ‘IRRIs’ can be identified as naturally occurring clay pellets (both quartz rich and quartz free) or naturally occurring opaque ferruginous inclusions (sometimes quartz rich).
Several other cases exist where macroscopic or even microscopic identification can be problematic. For particularly fine-grained fabrics, it can be difficult to distinguish quartz and feldspar precisely enough to record separate quantities in the hand specimen, and therefore they have sometimes been recorded together as QTZ/FEL. The distinction between flint and chert is equally difficult in both the hand specimen and in thin section. Nevertheless, most fragments in pottery made in Britain (particularly from the south-east) will be flint, while in imported wares (particularly those from the eastern Mediterranean) they are likely to be chert. For convenience the term flint is adopted for British material, and chert for imported wares, acknowledging that there will be exceptions to this guideline.
Size parameters. Inclusions have been estimated to the nearest micrometre and classified in accordance with the Wentworth scale (see Adams et al 1984, 3, table 1) and the terms ‘Silty’ = <0.0625mm and ‘Sandy’ = >0.0625mm have been used in the text. In some thin section descriptions finer gradations such as medium-grade sand are also employed. The term fine has been used when referring to mica measuring <0.1mm.
Sorting has been divided into ‘Ill’ and ‘Well’ sorted, referring to the size range of inclusions and to a certain extent their distribution within the clay body. While few Romano-British fabrics are ill-sorted in the manner of prehistoric pottery, many appear so when viewed at X20 magnification. Sherds with inclusions of a wide size range can be assumed to be ill-sorted, as can samples with a well-sorted groundmass containing sparse ill-sorted additional inclusions.
Frequency has been defined after FitzPatrick (1984, figs 5.2–5.8), and as illustrated on Fig 2 is defined as ‘Sparse’, ‘Common’ and ‘Abundant’ and recorded on the pro forma as such. In thin section, where greater accuracy is obtained, the term ‘Rare’ (<sparse) has also been employed.
Manufacture has been described as ‘Handmade’, ‘Wheelmade’ or ‘Other’ (described on the pro forma). Since the overwhelming majority of fabrics included here are wheelmade, manufacture is indicated in the text only for handmade or ‘other’ techniques.
Rounding parameters follow those defined by Pettijohn et al (1972), but as illustrated on Fig 3 have been simplified to ‘Rounded’, ‘Subrounded’ and 'Angular’, in keeping with the degree of accuracy visible at X20. For the same reason, shape is excluded for many inclusions <0.1mm when viewed in the hand specimen. When larger, the overwhelming majority of inclusions can be described as ‘subrounded’ and shape is therefore included only if distinctively ‘rounded’ or ‘angular’. Many inclusions (for example clay pellets or metamorphic inclusions) are subrounded to rounded by virtue of definition; similarly grog will almost invariably be angular. In these instances shape may be excluded from the written description. In thin section, where it is possible to observe shape more accurately, the full range of Pettijohn’s terms – as seen on Fig 3 – are employed.
Several other categories are included in the handbook which warrant explanation.
Source is intended to distinguish those industries which are known from kilns or production-like debris and those which are sourced on distributional grounds; where petrology is significant in assigning source, this has been noted.
The donor is acknowledged with thanks for each fabric type in the main body of the text.
Museums attempts to provide at least one museum or similar type of collection where a selection of the fabric type can be examined, although occasionally it has been necessary to cite museums that house only rare sherds. Foreign museums are given only for exceptional collections. Where available, information on archives and collections held by archaeological units is also listed.
References include the main bibliographic data. The extent and level of detail varies somewhat, for when a well-known synthesis is available pre-existing references are normally subsumed within it.
Chapters 2–7 of this handbook are organised by ware group and, apart from Gallo-Bellgic wares, are exclusively imported fabrics. Chapter 8, on Romano-British wares, is presented by county, within which fabrics are organised by industry. This has been done in order to facilitate description, to avoid repetition in the bibliography and to emphasise the relationship between different fabrics within an industry. For each fabric variables of colour, hardness, feel and fracture are described as General appearance, while Hand specimen refers to inclusions seen under the binocular microscope, as opposed to in Thin section. Other headings include Source, Donors, Museums and References. For industries with more than one entry, information may be combined and headings excluded from individual fabrics.
In contrast, the physical layout of the sherds is organised by ware and technological groupings (eg temper or surface treatment) regardless of source or industry, in order to facilitate the ‘matching’ of sherds. Correlation between the handbook and the physical collection is provided by Appendix 2.
The project was conducted using a Viglen Genie Professional 4SX25, operating on MS-DOS 6, and running Word Perfect 5.1. Information from the pro formas is stored in a dbaseIV database. Instructions for use of the database are housed with the computer and sherd collection. Variables relating to common name (and abbreviation), details of registration (including donor), provenance and thin section are computerised. The full text is also available on-line and can be searched for inclusion, colour, Munsell values and other fabric variables.
Several different mechanisms are in place to ensure that the collection does not deteriorate nor fossilise at our present state of knowledge. An Implementation Committee, comprising British Museum curatorial staff, an English Heritage representative and a member of the Study Group for Roman Pottery has been established in order to facilitate updating of the collection. This committee will address both the physical integrity of the collection as well as additions or alterations. Individuals or organisations wishing to contribute samples should contact the Department of Prehistoric & Romano-British Antiquities at the British Museum. As the subject develops, the Committee should consider new initiatives relating to the Collection, such as graphic facilities to incorporate and display computerised information on vessel form.
It is hoped that the national fabric collection will serve as a catalyst for regional collections, which will concentrate on local wares, including kiln material, as well as regional types found in the area. It will not be necessary to duplicate imported wares already represented in the NRFRC. Regional collections should be housed where a core collection or initiative exists, but a pragmatic and flexible approach must be taken, and different methods will suit different regions. Fabric collections can be augmented by a diverse range of facilities, such as museum collections, particularly of kiln groups. Alternative sources should be sought for the financing of these initiatives, including attachment to wider ceramic projects. These collections should allow the investigation of problematic local wares and kiln groups.
Adams, A E, MacKenzie, W S & Guildford, C, 1984 Atlas of sedimentary rocks under the microscope
Detsicas, A (ed), 1973 Current research in Romano-British coarse pottery, CBA Res Rep 10
Dragendorff, H, 1895 Terra-sigillata, Bonner Jahrbücher 96, 18-155
Dragendorff, H, 1896 Terra-sigillata, Bonner Jahrbücher 97, 45-163
Dressel, H, 1891Inscriptiones urbis Romae Latinae. Instrumentum domesticum, Corpus Inscriptionum Latinorum 15/1
FitzPatrick, E A, 1984 Micromorphology of soils
Freestone, I C, Johns, C, & Potter, T (eds), 1982 Current research in ceramics: thin section studies, British Museum Occ Pap 32
Fulford, M G, & Huddlestone, K, 1991 Current state of Romano-British pottery studies. A review for English Heritage, Engl Heritage Occ Pap 1
Fulford, M G, & Peacock, D P S, 1984 The Avenue du Président Habib Bourguiba, Salammbo: The pottery and other ceramic objects from the site, excavations at Carthage: the British Mission 1.2
Gillam, J P, 1939 Romano-British Derbyshire ware, Antiq J 19, 429–37
Gillam, J P, 1951 Dales ware: a distinctive Romano-British cooking-pot, Antiq J 31, 154–64
Gillam, J P, 1963 The coarse pottery, in Excavation at Mumrills Roman fort (K A Steer) 1958–60, Proc Soc Antiq Scotland 94, 113–29
Gillam, J P, 1970 Types of Roman coarse pottery vessels in northern Britain (3rd ed)
Kretz, R, 1983 Symbols for rock-forming minerals, American Mineralogist, 68, 277-9
Lais, R, 1943 Über des Herkunft römischer Amphoren aus Latenesiedlungen am Hoch- und Oberrhein, Germania 27, 50-2
Loughlin, N, 1977 Dales ware: a contribution to the study of Roman coarse pottery, in Pottery and commerce. Characterization and trade in Roman and later ceramics (ed D P S Peacock), 35–84
Middleton, A, & Freestone, I (eds), 1991 Recent developments in ceramic petrology, British Museum Occ Pap 81
Munsell soil color charts, 1992, New York (revised ed)
Orton, C, 1979 Dealing with pottery from a 600 acre urban site, in Pottery and the archaeologust (ed M Millett), Institute Arch Occ Publ 4, 61-71
Peacock, D P S, 1967 Romano-British pottery production in the Malvern district of Worcestershire, Trans Worcestershire Archaeol Soc 1, 15–28 (3rd ser)
Peacock, D P S, 1971 Roman amphorae in pre-Roman Britain, in The Iron Age and its hill-forts. Papers presented to Sir Mortimer Wheeler on the occasion of his eightieth year (eds M Jesson & D Hill), 161–88
Peacock, D P S, 1977a Ceramics in Roman and medieval archaeology, in Pottery and commerce. Characterization and trade in Roman and later ceramics (ed D P S Peacock), 21-33
Peacock, D P S, & Williams, D F, 1986 Amphorae and the Roman economy. An introductory guide
Pettijohn, F J, Potter, P E & Siever, R, 1972 Sand and sandstone, New York
Prehistoric Ceramics Research Group, 1995 The study of later prehistoric pottery: general policies and guidelines for analysis and publication, Occ Paper 1-2
Renfrew, A C, 1977 Production and exchange in early state societies, the evidence of pottery, in Pottery and early commerce. Characterisation and trade in Roman and later ceramics (ed D P S Peacock), 1-20
Rhodes, M, 1977 A pottery fabric type-series for London, Museums J, 76(4), 150-2
Rice, P M, 1987 Pottery analysis. A sourcebook, Chicago
Shepard, A, 1976 Ceramics for the archaeologist, Carnegie Institute of Washington Publ 609, Washington DC (8th reprint)
Swan, V G, 1984 The pottery kilns of Roman Britain, RCHM Suppl Ser 5
Tomber, R S, & Dore, J, 1996 A 'National Roman Fabric Collection', Britannia 27, 368-82
Webster, G (ed), 1976 Romano-British coarse pottery: a student's guide, CBA Res Rep 6 (3rd ed)
Williams, D F, 1977 The Romano-British black-burnished industry: an essay on characterization by heavy mineral analysis, in Pottery and early commerce. Characterization and trade in Roman and later Ceramics (ed D P S Peacock), 163–220
Williams, D F, 1983 Petrology of ceramics, in Petrology of archaeological artifacts (eds D R C Kempe & A P Harvey), 301-29
Young, C J, 1980 Guidelines for the processing and publication of Roman pottery from excavations, Directorate of Ancient Monuments and Historic Buildings Occ Pap 4