Method of fabrics research and classificationA.J. Nijboer & G.J.M. van Oortmerssen adapted from:
Pottery classifications: ceramics from Satricum and Lazio, Italy, 900 - 300 BC
1 NOTES ON FABRIC ANALYSIS A pottery fabric is described as a collective term for the internal constituents used in making pottery. These include the primary constituent such as clay or marl as well as any other material mixed into (e.g. temper) or processing technique (e.g. levigation) applied to the primary material to impart material characteristics such as hardness, porosity or thermal shock resistance (Adams & Adams, 1991: p. 356; see Appendix 1). Archaeologists have since long been aware of the material characteristics of the pottery but rather as a subjective criterion, such as the ware denominations used in this paper, for example impasto rosso and coarse ware. Specific wares like the Etruscan bucchero have intrigued scientist since long especially their manufacture (cf.. Del Vita, 1927). However, general, systematic classification of the pottery according their fabric was rarely promoted before the 1950’s when Shepard published her book Ceramics for the archaeologist (Shepard, 1956). Though Shepard herself did not use the term fabric as described above she definitely advanced the study of archaeological ceramics as matter. She described the physical properties of ceramics as including colour, hardness, texture, lustre, porosity, strength and other properties (Shepard, 1956: pp. 95-137). A systematic analysis of the above properties would advance an objective identification of ceramics as a group of material. Since Shepard fabric analysis included the above characteristics. For example, Vaughan stressed the use of similar macroscopic and microscopic characteristics of large groups of ceramics in order to maximise both the quantity and quality of data able to be retrieved from sherds (Vaughan, 1995: pp. 265-267). Fabric descriptions are useful for representing groups of material, offering a partial glimpse on ceramics, which need to be studied also from other perspectives such as typology. She also emphasises to concentrate on well-defined archaeological groups in order to systematise the observation of macroscopic material and technological features. Peacock & Williams (1986) and Tomber & Dore (1998) did publish monographs on such larger groups. Peacock and Williams placed Roman amphorae in a wider economic context but their data involved the description of the amphorae partly as fabric including a petrologic description. The recent handbook by Tomber & Dore (1998) concentrates on the characterisation of the fabric of Roman pottery found in North-West Europe, especially in Britain. Their aim was to provide a standard for the identification and description of Roman pottery types found in the United Kingdom but also elsewhere in Europe. The book presents numerous wares characterised by colour, sorting, inclusions, quantity, rounding, fracture and hardness. Eventually the research resulted in a physical reference collection with free access and stored at the British Museum "as a means for fabric verification, together with consistent and standard descriptions" of the sherds involved (Tomber & Dore, 1998: p. 2). The classification in fabrics presented in this article follows this line of research focussing on ceramic groups representing significant number of sherds. The specific methods employed are described in the next section. Two aspects remain to be discussed here, the notion of objectivity and an account on fabrics research in Italy. Vaughan warns that "the perception of progress in the concept of completely objective, codified and standardised studies of pottery, should be seen as the chimera it is" (Vaughan, 1995: p. 263). She recommends ‘pottery sense’, the skill to differentiate between, for example, "two undecorated sherds of Middle Minoan IIA and IIB conical cups without notable hesitation" (Vaughan, 1995: p. 262). Her plea is fully supported by our research since the fabric classification did require interpretation including their ware and type descriptions in order to demarcate multivocal, clustering characteristics for fabrics such as percentage of solid inclusions, which in itself is a clear property. However, one may wonder if the ancient potter ment a different fabric if the clay contained 11 or 9% solid inclusions (one of the demarcations in our fabric classification is based on less or more than 10% solid inclusions). Comparable complications regarding the domains of variability of individual attributes in fabrics research will be discussed in more detail in the next section. It needs however to be stressed that attribute clustering leads to identification of types (see appendix 1 for keywords and definitions). The fabric description has not been an objective per se. Its significance increases when combined with vessel typology and ware description. In Italy fabrics research has been promoted significantly by Cuomo di Caprio since the 1970’s, culminating in her book La ceramica in archeologia (1985). More recently various compilations and congress publications were published (cf. Failla, 1993; Olcese, 1995; Santoro Bianchi & Fabbri, 1997). The research presented concentrates on the role of archaeometry and on specific ceramic groups with a preference for Roman ceramics. Fabric analysis relating to the period 900 to 300 BC assessing ceramics from a major excavation derives from sites such as Acquarossa and Poggio Civitate (Wikander, 1993; Nijboer, 1998: pp. 161-171) while detailed ware descriptions have been published for Etruscan towns such as Caere (Cristofani, 1992; 1993) and Tarquinia (Chiaramonte Treré, 1999). One can detect a noticeable increase in the description of the material components of groups of ceramics from Italy but our approach of systematically describing the catalogued sherds from Satricum on a macro- and microscopic level culminating in a physical reference collection of the fabrics involved appears to be fairly unique so far.
2 FABRIC ANALYSIS AS ADOPTED BY GIA’S LABORATORY FOR CONSERVATION AND MATERIAL STUDIES 2.1 Procedures The fabric typology is based on variables of paste composition such as colour (before and after refiring under controlled conditions), inclusions and other attributes. Each characterising attribute analysed by the LCM is described below in detail. The procedure was as follows. After cataloguing the sherd (making an illustration, description and photograph), a small fragment was broken away using fine pinchers in order to make a fresh break. Fabric attributes were investigated mainly by analysing the fresh break. An old break would conceal certain attributes by soil and grime in the pores of the fabric. The individual features were than assessed using comparison charts for presence percentages and shape, Munsell Soil Colour Chart, hardness tests, sand ruler and binocular microscope. After identifying a fabric, a selection of sherds was refired at 600 ºC, 800 ºC and finally 1050 ºC in an electric, oxidising kiln in order to establish firing characteristics as well as the final colour of the fabric since the colour of the excavated sherd is often affected by less controlled firing in antiquity, usage and post-depositional changes. Refiring sherds eliminated greatly the colour variables of a fabric. Each attribute received a code leading, at first glance to a complicated fabric code. Much of the description of the fabric is enclosed in this code, the key of which will be presented further on in this section. An advantage of codes is that is becomes clear if fabrics are related or not and if they belong to the same family. Eventually at least two thin sections were made per fabric for a petrologic description. Since the campaigns of 1997 fabric analysis was systematically performed by Gert van Oortmerssen. This implies that the majority of the ceramics catalogued before 1997 were not subjected to a standardised analysis of their fabric. The existing pottery typology became thus enhanced with a fabric typology till the last Satricum campaign of 2001. The present publication presents mainly the technical information regarding the various classifications of the impasto and coarse ware pottery. In future publications we will present in more detail artefact groups such as ceramic building materials, bucchero and figulina referring to the present paper for the methods employed. The fabric classification led to the establishment of a physical fabric reference collection for both Satricum and the wider region. This collection contains the most significant sherds for the specific fabric. In addition the fabric reference collection gives per fabric an impression of the types and wares of pottery involved.
2.2 Method of fabric research and codes The colour of the ceramics appeared to be the most distinctive feature, especially after refiring. It is therefore placed at the beginning of the code following the abbreviation of the site. Refiring of the sherds was carried out under standardised conditions: oxidising at 600 or 800 oC for one hour precisely. Quite a few fabrics would eventually merge by detailed shifting of the variables of the attributes such as colour, predominance of (a) specific inclusion(s) and total percentage of inclusions. Colours were described using Munsell soil colour chart (1994, revised edition). The main colour types of fabrics are:
Other decisive material variables examined for the fabric classification involve mainly the inclusions in the clay and some specific qualities regarding hardness, porosity and fracture. These variables are:
On the basis of the variables presented above and in figure 1 we will now introduce the key for the codes employed. The following order for each fabric code is maintained:
The origin of the sherds is expressed by three capitals, cf. SAT (Satricum), SEG (Segni) or LAV (Lanuvium):
Undercast (small) characters in the code are used when:
An asterix, ‘*’, behind codes for types of inclusions is used in cases where dominant inclusions, sorting or particle size can only be determined under the microscope. Refiring is carried out under standardised conditions: oxidising at 600 or 800°C for one hour precisely. ‘Related to’ means that two or more fabrics are similar in material characteristics though there is a difference in one or two of the attributes of the fabrics. Quite a few fabrics can be grouped as fabric families. Colours are described using Munsell Soil Colour Chart (1994, revised edition).
2.3 Examples of codes and fabric descriptions Here we provide a few examples of fabric descriptions: SAT II.AD*.ms(2).ab, Augite/hardness 3-4 SAT I.ADe.ps-vps(1-4).a LAV I.ek.ps(1-3).cd SEG III.=.ws(4).d, powdery surface
3 VARIABLES 3.1 Colour The colour of the excavated pottery is the most characteristic feature of the fabric description. The examined fabrics demonstrate that in time there is a gradual shift from red – and red to orange firing ceramics dated mainly to the 9th till the 6th century BC to orange – and pale firing ceramics dated predominantly to the 7th till 4th century BC. This shift is primarily due to the change in manufacturing techniques of the ancient ceramic industry and the introduction of kilns with separated combustion and firing chambers during the 7th and 6th century BC (Nijboer, 1998: pp. 73-196). This led to a standardisation not just in the ceramic repertoire but also in a decrease in colour variables of the pottery. However, the colour as it presents itself after excavation is the result of various processes such as manufacturing procedures, firing, use in Antiquity and post-depositional changes. This results in numerous colour variables of the excavated ceramics, which are not directly related to the standard colour of the fired clay in an oxidising atmosphere. In order to eliminate some of these variables, the pottery was refired at 600, 800 and 1050 °C. Appendix 4 presents the colour variables as they have been measured after excavation and after each successive refiring programme. The documented Munsell measurements in Appendix 4 demonstrate that the colour variables decrease with each refiring step. Thus it became simpler to ascribe ceramics to a specific colour range/type. It was noticed that the colour of the oxidised zone just underneath the surface of the excavated sherd, if present, was a good indication of the colour obtained after refiring. A colour brown as in the ware group ‘common brown impasto’ (section 3.2), does not exist as a colour type in the fabric description since after refiring in an oxidising atmosphere, it became red or red to orange.
3.2 Inclusions The inclusions present in the fabrics from Latium Vetus may contain the following inclusions ((v) = volcanic). Rockforming minerals
Rockfragments
Other
Intentional
Unknown
Organic
The minerals are identified by using mineral descriptions, thin sections and physical mineral references collected in the region, occasionally identified by XRD.
3.3 Sorting of inclusions
Values on the sand ruler (with 10 fractions):
3.4 Percentage of inclusions
3.5 Other characteristics
Characteristic placed between brackets in the fabric code indicates that it is not found in every, single sherd, but significant for the fabric as a whole. If a characteristic is indicative for a fabric but not listed in the fabric code by emblem it is described under ‘features by eye’ between brackets, for example: powdery surface.
4 COMMENTS As mentioned above most of the attributes (3.1 to 3.5) are assigned after assessing a range of variables for each characteristic. A fabric type as listed is however consistent because the same overall set of variables is assessed and any given type cannot completely overlap with the definition of any other type. A combination of all the attributes including their variables leads to distinct, individual fabric assignments because combined attribute variables exclude each other. For example some sherds assigned to the red family may overlap in colour with sherds assigned to the red to orange fabric. However colour is only one of the five attributes assessed and it is impossible that variables of the other four characteristics coincide as well. One may detect however from the code those fabrics, which only differentiate in one or two of the attributes. For example, the most dominant fabrics at Satricum are:
From the fabric code one can deduce that both fabrics are very similar demonstrating that they belong to the same fabric family and only differ somewhat from each other in sorting and percentage of inclusions. Furthermore a final assessment of each fabric involves also the ware and pottery typology. Appendix 3 gives a complete description of the specific fabric containing the earliest architectural terracottas recovered at the Satricum and dated around 600 BC. A page with figures illustrates the architectural terracottas as well as the vessel forms attributed to the same fabric.
5 RESULTS: THE FABRIC CLASSIFICATION All in all 53 fabrics were identified in the past 5 years (appendix 2). This section will list 13 fabrics in order of prevalence at Satricum. The order derives from the number of sherds analysed per fabric. This number gives an indication of the dominance of fabrics excavated by the Groningen Institute of Archaeology at Satricum. The 40 fabrics (53-13 = 40) not discussed in detail will be presented at a later stage or consist of only a few sherds, not enough to form a consistent group. Some of these sherds might be actually imports.
1 SAT I/(II).AD*.ms-vps(1-4).a, variety of characteristics/(slightly) gritty 1634 sherds attributed to this fabric. Wares: impasto bruno, impasto rosso and a few sherds coarse ware. The fabric is locally made. Present as fabric in 7th century BC pottery kiln (Nijboer, 1998: pp. 115-121) but also in the third production site at Satricum, dated to the 5th/4th century BC. Vessel forms: jar, storage jar, miniature jar, miniature cup, tile, bowl, large bowl, lid, plate, holmos, fornello, spindle-whorl, spool, loomweight and mug. Date: Late 9th till 5th/4th centuries BC, most of which dated to the 8th, 7th century BC (fabric present in some of the huts). Limited numbers of coarse wares and Archaic/5th century BC vessels though the use of the fabric definitely continued into the 5th/4th centuries BC. The few coarse ware sherds present may be due to the assessment of the variables. A typical coarse ware is indeed related (SAT I.AD*E.ms-ps(1-4).a, fine gritty (with considerable more augite). Other forms executed in this fabric are fragments of rooftiles, a terracottafragment as well as a temple model, all dated to the 5th century BC. The fabric also includes a jar which can be considered as one of the oldest measures of capacity in Italy (Nijboer, 1998: fig. 73, jar 1). This indicates that this measure was probably made locally and that the knowledge for making such measures of capacity was available at Satricum. Thin sections are available. Fabric present in some of the huts; Related to the following fabrics:
2 SAT I/(II).AD*.ms-ps(1-4).b, variety of characteristics 676 sherds attributed to this fabric. The difference between this fabric and the previous one concerns mainly the sorting and amount of inclusions (a. indicates more than 20% solid inclusions while b. indicates 10 to 20% solid inclusions). Some sherds assigned to this fabric derive from 7th century pottery kiln. Wares: impasto bruno, impasto rosso and a few sherds coarse ware. Vessel forms: jar, storage jar, tile, lid, holmos, fornello, sieve, spindle-whorl, spool, cup, bowl and amphora. Date: late 9th till 6th century BC. Thin sections are available. Related to the following fabrics:
3 SAT I.=.ws*(1-4*).d,((large)FeMn nodules) 585 sherds attributed to this fabric that has a very low percentage of visible inclusions (< 5%), which appear well sorted though not under the microscope. This fabric is suitable for fine wares reflected in the vessel forms including many decorated (incised and relief decorations) tablewares, slipped vessels sometimes burnished to lustre. The fabric includes the typical Satricum anforiskoi, kantharoi and bowls identified by Beijer (Beijer, 1991R) as well as a vessel type related to the ‘Phoenician’ plates (classified by Beijer as Bowl V-6), similar to those found in the Western Mediterranean (Peserico, 1998). Wares: impasto bruno, impasto rosso and a few sherds coarse ware. Vessel forms: carenated bowl, bowl, plate, jar, cup, amphora, strainer, stand, spool, spindle-whorl, dolium and jug (‘orciolo’). Date: 9th till 6th centuries BC. Thin sections are available. Related to the following fabrics:
4 SAT I.ad*eq.ps-vps*(1-4).c, (large) FeMn nodules 495 sherds attributed to this fabric, which is related to the above fabrics. This fine fabric contains a low percentage of inclusions (5 to 10%), which vary in size. It consists predominantly of tablewares. Wares: impasto bruno, impasto rosso and a few sherds coarse ware. Vessel forms: miniature plate, miniature bowl, miniature cup, spool, spindle-whorl, loomweight, jar, storage jar, amphora, fornello, sieve, plate, holmos, jar, lid, bowl and cup. Fabric present in some of the huts. Date: late 9th till 6th/5th centuries BC. Thin sections are available. Related to the following fabrics:
In the above 4 fabrics, coarse wares are included but they only occur as intrusion. The bulk of the ceramics involved are impasto bruno and impasto rosso wares. This fabric family is a typical fabric family of the Iron Age and Orientalising period. The few coarse ware sherds can be accounted for since they are related to typical coarse wares such as SAT I.AD*E.ms-ps(1-4).a, fine gritty.
5 SAT II.AD*.ws(3-4).a, ((black) organic) Fabric with 246 sherds attributed. It is related to fabric 8 though with finer texture and well sorted inclusions (< 250). The fabric contains mainly tablewares but also some 5th, 4th century BC large jars. From the assessment of a fresh break there is no distinction possible between the 7th century BC impasto rosso plates/bowls and the 5th, 4th century BC vessels. The distinction in date becomes however obvious when one considers the surface treatment as well as the typology of the vessel forms. Wares: orange coarse wares, impasto rosso, late, grey bucchero. Vessel forms: bowl, cup, jar, jug, amphoriskos and plate. Date: 7th century BC and 5th, 4th centuries BC. Related to the following fabrics:
Thus fabric 5 is related to SAT II.AD*.ws(4).abc, which consists of bucchero sherds dated to the 7th century BC, with vessel forms calice, kantharos and amphora. This substantiates the close relation between the local wares impasto rosso and bucchero during the 7th century BC at Satricum. SAT II.AD*.ws(4).abc has 46 sherds attributed, which coloured orange after refiring in an oxidising atmosphere..
6 SAT I.AD*K*.vps*(1-4*).ab,occ.(large) FeMn/occ.(small) augite Fabric containing domestic pottery as well as the oldest architectural terracottas at the site (157 sherds attributed). Full fabric description in appendix 3 as example of the administration of an individual fabric. Wares: impasto rosso and impasto bruno. Vessel forms: tile, cover tile, acroteria, small terracotta head (probably decorating an architectural terracotta), bowl, calice, cooking stand, stand, storage jar, large bowl (teglia), miniature jar, jar, lid and plate. Date: late 7th, early 6th century BC. Thin sections available. Related to the following fabrics:
7 SAT I.AD*E.ms-ps(1-4).a, fine gritty Fabric with 156 sherds attributed and related to fabric 10 though with finer mineral inclusions. Some sherds derive from 7th, others from late 6th century pottery kiln. Distinction between wares is clear judging from the whole appearance of the sherds though less clear from a fabric point of view, which analyses the appearance of the surface of a fresh break. Wares: mainly coarse ware, some impasto rosso and impasto bruno. Vessel forms: Jar, storage jar, stand, lid, mug, bowl (teglia), spool, cooking stand and tile. Date: 8th till 5th century BC. Thin sections available. Related to the following fabrics:
8 SAT II.ADe.ms-vps(1-4).ab , variety of characteristics 125 sherds attributed to this fabric. Fabric with restricted repertoire mainly occurring in Votive Deposit II, La Fornace survey and Southwest necropolis (Bouma, 1996; Nijboer et al., 1995; Gnade, 1992). Wares: coarse wares. Vessel forms: jar, large bowl (teglia), votive terracotta and lid. Date: late 6th, 5th, 4th centuries BC. Related to the following fabrics:
9 SAT II.AD*.ms-ps(1-4).a*, (augite)/((small) black stains) Fabric with 119 sherds attributed. Wares: impasto (orange). Vessel forms: jar, bowl, large bowl (teglia) and antefix. Comparable to Knoop Fabric 2 described as ‘Campanian’ (Knoop, 1987: pp. 228-229). Date: 6th century BC. Thin sections available. Not related to other Satricum fabrics.
10 SAT I.AD*EQ.ps-vps(1-4).ab, medium-coarse gritty/(hardness+) Fabric with 101 sherds attributed including the red architectural terracottas dated to the 6th century BC among which wasters from the late Archaic kiln (Nijboer, 1998: pp. 121-129). Fabric contains 15 to 25% solid inclusions and is therefore suitable for the manufacture of the larger, coarse vessel types as testified by the vessel forms. Wares: impasto (red) and coarse wares. Vessel forms: Storage jar, plate (large), tile, cover tile, lid, jar and bowl (teglia). Date: 6th century BC. Thin sections available. Related to the following fabrics:
11 SAT I.AD.ws(1+3/4).a, ws tiny white dots/(large) FeMn nodules Red impasto with well sorted inclusions (67 sherds attributed). Wares: impasto rosso Vessel forms: amphora, storage jar, jar, bowl (teglia) and plate. Date: 7th till 5th centuries BC. Thin sections available. Not related directly to other Satricum fabrics.
12 SAT II/III.E.ps-vps(1-4).a, coarse gritty/Leucite-lava & Leucite-tuff Fabric belonging to the Late Archaic Fabric Family of pale impasto containing considerable amounts of augite (29 sherds of this specific fabric attributed). Wares: Pale impasto (impasto chiaro). Vessel forms: tile, storage jar, loomweight, large bowl (teglia), antefix (Juno Sospita, Satyr and Menade), fragments of life-sized terracotta statues and wasters from Late Archaic Kiln in A13-14 (Nijboer, 1998: pp. 121-129). Date: late 6th, 5th centuries BC. Thin sections available. Related to the following fabrics:
13 SAT III.E.ms(1-3).ab, fine gritty Fabric belonging to the Late Archaic Fabric Family of pale impasto containing considerable amounts of augite (25 sherds attributed). Wares: Pale impasto. Vessel forms: bowl, loomweight, votive arula with naked four-winged male figure, tile and large bowl (teglia; among which those found in the Late Archaic kiln in A13-14: Nijboer, 1998: pp. 121-129). Date: late 6th, 5th centuries BC. Thin sections available. Related to the following fabrics:
Keywords and definitions
List of described fabrics present in the reference collection
FAMILY I
FAMILY I/II
FAMILY II
FAMILY II/III
FAMILY III
Example of a full fabric description
SATRICUM FAMILY I Fabric SAT I.AD*K*.vps*(1-4*).ab, occ. (large) FeMn/occ. (small) augite
Properties:
Related to:
Number of sherds in reference collection: 107 Thin sections: 5
Colour variables of ceramics from Satricum, after excavation and after each successive refiring programme. Legend: bold = dominant colours; italic = interaction with other fabric families in case of dominant colours.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
