As for metal and glass, the study of ceramics on its own can provide the substance for a book – and indeed it has. For example, a comprehensive publication by Prudence Rice (1987) has considered the full complexity of ceramic technology; a volume edited by Freestone and Gaimster (1997) provides a broad and informative consideration, based on short case studies, of the range of ancient ceramic technology. Even though this chapter focuses on the techniques of ceramic production, and the technology involved in forming and firing pottery, there are associated realms of study which should not be forgotten, and provide a more holistic approach. Certainly the ‘evolution’ of ceramic technology is undeniable, but there are many exceptions to what some people might call progress. There are certain assumptions in the study of pottery from an evolutionary perspective. Perhaps the most significant is the assumption that raw materials are chosen because they impart specific properties to the clays when they are being formed into a pot, when they are fired and when being used in one or more contexts. Of course in many cases this is true – simply because the behaviour of the pottery has been observed and the technology has developed on the strength of those observations. Related to the more deterministic area of pottery research is that of ceramic ecology, the study of ceramic production in relation to the exploitation of raw materials in the environment in which it occurs (Matson 1965; Kolb 1989).
However, as White (1972) and Wright (1986) have discussed, the interaction of style and technology is more complex than this: ‘there is nothing inevitable about the acceptance of a new technology’ (Wright 1986: 17); among other social groups, potters may have played an active role in creating stylistic changes. In addition to this, there is the study of pottery production in the context of the ways in which craft groups may determine the materials added to the pots that do not necessarily play a significant role in their performance but are determined by what was used or demanded by potters and consumers – so-called cultural factors (Wright 1986; Kilikoglou et al. 1998). Bishop (1992: 169) has stressed how important it is to keep ‘central the archaeological problem under investigation … thereby avoiding the centripetal tendency for methodological rather than archaeological statement, whether investigating changes in technology, production strategies, or ceramic exchange’. It is clear that the full range of potential sources of variation which produce the archaeological record should be considered. Tite (1999) has discussed a complex range of parameters which have led to the various modes of pottery production, distribution and consumption. Rather than discussing these parameters in isolation he has shown how the physical sciences can provide the means of investigating some of these questions; at the same time he accepts that ‘sociopolitical and cultural–ideological factors are clearly of very considerable importance in explaining technological choice’ (ibid.: 225). Something which often forms part of the archaeological investigation of ceramic composition is whether ‘workshop compositions’ of clays used can be identified. An analytical study of third-and second-millennia BC pottery from Iran (Henrickson and Blackman 1992) has revealed links between typological characteristics such as vessel form, motif repertoire and decorative style, and the use of restricted ceramic compositions. It can be argued that indeed the products of workshop traditions have been identified – though there is still a further possible explanation: that for some (political) reason the raw material supply was controlled resulting in the same raw materials being used in a number of (nearby) workshops.
Part I of this chapter is a consideration of how clays originated, how clays were prepared for pottery production, how pottery was made and decorated, including the equipment needed and aspects associated with kiln use. Part II provides in-depth case studies: the changing mode of Iron Age pottery production in southern England; early medieval pottery production and the growth of towns; the manufacture of Chinese celadon; and Ottoman Iznik and the court patronage. All of these case studies are a reflection of the social and economic contexts in which the pottery was made, the level of archaeological research involved and the ways in which scientific investigations have been carried out.
Although common clays are complex and need to be prepared before pots can be made from them, one of their more important properties, which people often take for granted, is their ability to absorb water. This allows clays to become plastic so that they can be shaped – the property of plasticity. Not all clays exhibit the same plasticity; this variation can be linked directly to their physical structure. Since plasticity is a critical property of clays, their physical structure and the ways in which they were formed are clearly worth considering in some detail since it is these which contributed to the property of plasticity.
Clays as materials can be defined in a range of ways. They are complex and are formed in many different contexts which, in turn, lead to their different physical (and chemical) properties. It is the existence of this wide range of clay types which consequently leads to the manufacture of pottery of a correspondingly wide range of colours and properties. The potter’s intervention in preparing and modifying the clays, including transformation by fire, also has an important impact on the finished product.
Clays can be defined in a variety of ways (see Section 4.2.3), but in order to define this most important component of pottery, it is necessary to start from basics: the composition of the earth’s crust. It is the earth’s crust which ultimately provides the raw materials from which clays are formed. The commonest elements in the earth’s crust (expressed as oxides) are silicon dioxide – Si02 (60.1%) and aluminium oxide – A1203 (15.6%) with the balance of oxides, including the oxides of iron, magnesia, calcium, sodium, potassium, titanium and phosphorus being present at less than 5.1%.
Minerals are one of the principal components of clays. These are formed from a combination of oxides with a range of other compounds which occur in the earth’s crust. A mineral is defined as any one of a number naturally occurring solid inorganic substances with a characteristic regularly ordered crystalline form. Different mineral types have different characteristics in addition to their crystalline form, which set them apart. These include colour, fracture, lustre (the way in which light is reflected from the surface of the mineral), hardness and specific gravity. The characteristic of fracture, for example, leads to crystal shapes which distinguish, in a very general way, between one family of minerals, quartz, feldspars and olivines and a second family, the pyroxenes and amphiboles. Following weathering and transportation (see Section 4.2.3) the first family are a blocky, squarish shape with equant forms, whereas the second family of minerals retain a sticklike, longer and thinner shape. Quartz tends to split (cleave) at its edges, pyroxenes split along their long axes (following cleavage planes) – a third family of mineral, the micas, cleave as sheets along the planes of their structures. In the chapter on metals the ore minerals are described (Section 5.2); those of relevance to the discussion here are rock-forming minerals because all clays are formed from rocks. Thirty-nine per cent of the surface rock-forming minerals is composed of feldspars. Feldspars are a combination of silica (60.1% of the earth’s crust) and other elements which form silicates. The mineral quartz forms 28% of the surface rock-forming minerals, clay minerals and micas 18% and ferro-magnesian silicates 2%; 9% are carbonates.
What bearing does the occurrence of these minerals have on the use of clay by a potter? The clays used by potters are essentially the result of sedimentation, but the sediments themselves have been formed by the weathering of rocks (igneous, metamorphic) which are composed of minerals. Since igneous rocks basically give rise to the formation of metamorphic and sedimentary rocks, it is worth considering them in some detail, because the properties of clays are fundamentally linked to them.
Igneous rocks are formed either by volcanic eruptions (produced by volcanos) or plutonic extrusions (formed deep in the earth and cooled slowly). Examples of fine-grained volcanic rocks are obsidian (a natural glass; see Section 6.6), basalts and pumice; examples of coarse-grained plutonic rocks are granites and dolorites. Igneous rocks can also be classified according to a scale of acidity (high silica) and basicity (low silica). Thus acid rocks like granites have a pale colour, are coarsely textured (because they have cooled slowly) and have a low specific gravity; basic rocks like basalts have finer textures (because they have cooled faster) are darker in colour and have a high specific gravity. The greater weight and (darker) colour of basic rocks derive from the large quantities of mafic (magnesia and iron-rich) minerals.
As mentioned above, the most common rock-forming minerals are those which contain silica. The rock-forming minerals primarily with silicates can be graded according to their acidity or basicity, from the most acid minerals (micas), through quartz and feldspars to the ultra basic pyroxenes and olivines. Of course, rocks are made up of a variety of minerals.
Igneous rocks are the basis for metamorphic rocks (transformed by heat and/or pressure); sedimentary rocks are the result of the transport and deposition of igneous or metamorphic rocks. Some sedimentary rocks are formed by the compaction of millions of the shells (the ‘carapace’ of organisms that once inhabited them) and of the skeletons of diatoms – tiny single-celled algae which have skeletons impregnated with silica.
In describing the physical properties of igneous rocks and the minerals that compose them, their texture and acidity were mentioned. These properties will determine directly the susceptibility to weathering of igneous and metamorphic rocks (see Rice 1987: Table 2.3) and thus the kinds of clays that are formed from them, and their firing properties. However, having described the ways in which rocks and minerals can be classified, when we consider the most significant rocks for the formation of clays, it is the breakdown of those silicate rocks which contain significant aluminia, such as micas and some feldspars, that we are concerned with here. Feldspars contain a combination of silica and aluminia, and the balance of their compositions leads to a corresponding range of feldspar compositional types produced mainly by varying proportions of potassium, sodium and calcium. Thus feldspars can be classified into alkali (potassium) feldspars (orthoclase and microcline) and soda-lime (plagioclase) types: albite – is the sodic form and anorthite – the calcic form, depending on the proportion of these components. There are also plagioclase feldspars with intermediate compositions and different names (Velde and Druc 1999: 21–2).
The weathering and decomposition of rocks and minerals to produce clays involves mechanical, chemical and biochemical agents. Two processes are generally involved: fragmentation by wind, water and glacial ice and the chemical reaction of minerals with solutions of water (hydrolysis) to produce new minerals, sometimes involving weak acids. The processes of weathering (like any chemical reaction) are affected by temperature (extremes of temperature as well as the mean temperatures) and by rainfall, perhaps in the form of flooding. The biological agents that can play a part in breaking down rocks include bacteria, algae and rootlets. Extreme forms of chemical weathering remove elements from the rocks according to the solubility of the components – sodium first, followed by potassium, calcium and magnesia – which leaves relatively insoluble iron, aluminium and silicon. This kind of weathering occurs in humid tropical regions.
Clays can be defined in several different ways, including the following features:
(I) depositional situation;
(II) particle sizes;
(III) chemical composition and structure;
(IV) mineralogy.
These definitions will be summarised here and the principles are described in each case, but it should be borne in mind that clays are extremely complex and a satisfactory agreement between researchers on how best to define them has not necessarily been reached in all instances.
These various ways of defining clays can be used to attempt to ‘label’ them. However people in the past who selected clays for pottery making would have done so with the experience of how the clays behaved when worked, especially their plasticity/the extent to which they absorbed water, how the clays were changed when fired (their shrinkage and thermal shock) and how different pottery colours were achieved when fired under a variety of conditions. In any case our labels obviously had no significance to past societies, but the behaviour of clays during the production process led to an enormously wide range of pottery types; clearly the development of firing technology and the mixing of clays both had their important roles to play. These (physical) definitions can be related to visual characteristics of pottery, including the ways in which the clays have been transformed in the fire and with the addition of other materials.
Primary (or residual) deposits of clay are those which have remained in more or less the same situation as the parent rock (such as granite, basalt, limestone and shale) from which they were derived as a result of various weathering processes described above. Sometimes when the parent rock is incompletely weathered these residual clays contain fragments of the parent rock such as feldspar, mica and quartz.
Secondary (sedimentary or transported) clay deposits are those which have been transported some distance from the parent rock by natural agencies such as wind, glaciation or streams. The largest sedimentary clays are those carried down to the coast and deposited as marine deposits. Because they are generally finer-grained, secondary clays they can be more homogeneous and can contain up to 10% organic components. Glacial clays generally contain a relatively high proportion of impurities and tend to be unsorted and are suitable to make earthenware from.
Different formation processes of clays lead to different particle sizes. These can be measured using a technique called granulometry (the measurement of grain size); it is generally agreed that clay contains particles of predominantly less than 2 micrometers (Rice 1987: 38, Figure 2.2). Because clays consist of such small particles they behave like colloids. A colloid is a mixture consisting of particles of one component suspended in a continuous phase of another which has the properties between those of a solution and a fine suspension. An example of an organic colloid is blood where haemaglobin (for example) is suspended in blood serum. This definition of clays, as colloids, provides a useful insight into how they behave when they are manipulated and worked, though not all particles in clays are a colloidal-size. According to several different (institutional) definitions of clays they contain the finest particle sizes, with silts, sands and gravels containing progressively larger particles. Soil maps frequently include silty clays and clayey loams, but fine-textured plastic clays are those deposited in lakes, streams and estuaries. For a soil to exhibit plasticity it should contain a minimum of 15% of fine particles.
Just as clays have distinctive properties which enable them to be made into pottery because they are colloids, their chemical compositions (and structures) are equally important. Clays are essentially composed of the two commonest oxides in the earth’s crust, silica (SiO2) and aluminia (A12O3): it is no coincidence that these also happen to be the most resistant to weathering. When combined with water (H2O) these oxides form what are generally described as hydrous aluminium silicates; most clays are composed of this substance, and obviously the existence of different clay types is connected to the different proportions of silica, aluminia and water they contain. The (atomic) structure of different clay types varies to the extent that the ratio of silica to aluminia can vary from 1:1 to 4:1, or higher, and the water component can vary between c. 13% and 35%. In addition, other oxides introduced by the presence of various minerals determine their chemical compositions. Iron and aluminium also combine with water, especially in tropical and subtropical zones, to produce iron and aluminium hydrous oxide clays, sometimes mixed with silicate clays.
These structural ‘units’, silica, aluminia and water, in fact are typically arranged in sheets, but not as discrete oxides. The ‘water’ component breaks down into oxygen and hydroxyl (OH–) components in clays and combines with silicon and aluminia in different ways, producing essentially two ‘building blocks’: (1) silicon combined with oxygen atoms and (2) aluminium combined with oxygen atoms or hydroxyl groups (see Figure 4.1). The relative number of these groups is determined by the kind of clays involved; as with any material the charged atoms or groups of atoms (ions) which constitute clays are held together with electrical charges. Silicon combines with oxygen atoms to produce tetrahedrons consisting of a central silicon atom combined with four oxygen atoms equally spaced around it; aluminium atoms combine with two oxygen and four hydroxyl ions forming octahedrons. The silicon-oxygen tetrahedra (the basic building blocks of silicate rocks) form sheets in clays which are typically composed of groups of tetrahedra with shared oxygens at their corners, producing hexagonal rings (Rice 1987, Figure 2.4b, c). In a similar way, sheets of octahedra are formed from aluminium, oxygen and hydroxyl groups. However, magnesia and iron ions (e.g. Mg2+, Fe2+, Fe3+), among others, can replace ions of silicon and aluminium and this compositional range leads to a corresponding range of 50 clay mineral types. Trace components, such as titanium, calcium and sodium may also bond with clays.
Figure 4.1 Some of the structural units found in clays: (a) silica and oxygen tetrahedron; (b) and (c) hexagonal arrangements of silica and oxygen tetrahedra; (d) aluminium, oxygen and hydroxyl octahedron.
Silicate clay minerals are often composed of sandwiches of alternating silica tetrahedra and aluminia-rich octahedra which are weakly-bound together. This clay structure produces platelet shaped particles and result in easy cleavage (see Section 4.2.3).
The use of optical petrology and X-Ray diffraction (see Section 2.2) has shown that clays basically consist of fine crystals with an internal structure such as silica and aluminia-rich sheets. Having said this, the existence of ‘transitional’ clays illustrates the point that the definition and classification of clays is far from simple and there are still areas where no consensus of opinion has been reached. In addition, since in attempting to reconstruct past systems of pottery production we may want to relate the source of particular clays used to the fired result, it is worth bearing in mind that by firing the clays some of the mineralogical structure may be destroyed.
Most clay minerals or mineral groups can be classified as layered silicates (also known as phyllosilicates) (IV-1). Some clay minerals have a chain or lath structure (IV-2); a small number of clay minerals (allophane minerals) are amorphous to X-Ray diffraction.
Differences in the arrangement of the silicon- and aluminium-rich layers as well as the substitution of aluminium by various other groups, including those rich in potassium and magnesia, produces the variation which allows for a tripartite subdivision of layered silicate clays.
These are two-layered structures, one each of silica tetrahedra and aluminia octahedra which are bonded relatively strongly. Kaolinite clays are mainly composed of the clay mineral kaolin. Kaolin results from the advanced weathering of rocks like acid granitic rocks (such as pegmatites and micaceous schists). Kaolin is normally high in aluminia – one of the characteristics of Chinese porcelain – and often occurs in the ratio of 2:1 with silica, combined with water. It normally occurs as flat hexagonal plates. Kaolins occur widely in temperate and tropical zones; the familiar conical hills of kaolin (‘china clay’) at St Austell in Cornwall are a memorable sight. These hills result from hydrothermal alteration of the granites which are characteristic of the area.
These clays are relatively low in plasticity, they are coarse and contain impurities including fragments of parent rock; the better sorted sedimentary kaolins tend to be relatively free of impurities, such as iron, which would colour the clay; the result is that they are white when fired. China clays have a high firing temperature (refractory), are low in impurities and, because of their relatively large particle size, are low in plasticity. They have a relatively low shrinkage when they are dry.
Hallyosite clays have developed from volcanic ash as recently as 4000 years ago (Hay 1960). They tend to be formed by hydrothermal action. Although they too have a two-layered structure, irregularities in the structure lead to more water in the mineral structure than kaolinite clays.
These clays have a three-layered structure consisting of two sheets of silica tetrahedra separated by a single layer of aluminia octahedra, again bonded loosely. Water molecules and a variety of atoms can easily penetrate the spaces between the unit layers which makes them expand, cleave apart and adsorb additional ions with ease; these water molecules are known as interlayer lattice water to distinguish them from chemically-bound water or the very weakly-bound water which contributes mainly to the plasticity of the clay (see Section 4.3.2 below). Smectites are formed mainly by the alteration of basic rocks and minerals which are high in calcium, magnesia and iron; examples are basalts or the decomposition of volcanic ash. Smectites are a relatively common component in arid regions and in recent sediments, but are not as highly weathered as kaolinites. If smectites are leached when high rainfall, high temperature and good drainage occurs, bases (such as magnesia, sodium and potassium) will be removed and this will lead to the formation of kaolinites. When compared to kaolinitic clays, smectites have a higher ratio of silica to aluminia (around 4:1), they contain more alkali metal ions such as sodium (Na+) and potassium (K+) and, because of the relatively lower level of aluminia, the clays are less refractory and melt/are fired at lower temperatures.
The most common member of the smectite group is montmorillonite. A number of ionic substitutions can occur within the clay: for example, if Al3+ is replaced by Mg2+ saponite is produced, or Al3+ by Fe3+ nontronite is produced. Smectite particles are thin and platey, but can be considerably smaller than kaolinite particles (0.05 μm-l μm, rather than 0.6 (μm-1 μm in diameter). The small particle size means that smectites are often plastic and sticky; their tendency to absorb water means that they display a high degree of shrinkage and have a tendency to crack when drying. Because they are able to adsorb other (colorant) ions they can be used as coloured slips.
These also have a three-layered structure; they were named in 1937 after the state of Illinois in the USA, and are similar to smectites. About 15% of the silicon is replaced by aluminium in these clays, so they melt at even lower levels than smectite clays. At the same time, a charge deficiency, primarily in the outer silica layers of the unit structure, is balanced mainly by potassium ions (K+), and in addition Ca2+, Mg2+ and H+ ions. Because the deficiency is close to the surface, rather than in the interior as with smectites, illite clays are non-expanding. Illite minerals, like smectite minerals, are poorly-defined flakes with diameters ranging from 0.1 μm to 0.3 μm. These clays are also good for making coloured slips, such as used on Arretine/Samian ware and Greek black-figure ware.
Illite clays are produced in a variety of environments. They can form in offshore or deep alkaline marine environments by the alteration of kaolinites and smectites in the presence of calcium, potassium and magnesia and the absence of a leaching solution. These clay minerals can form by diagenesis. Over thousands of years illites may alter and become smectites.
These are fibrous or lath-structured clay minerals which are found particularly in arid and desert regions; they can be weathered products of basalts. The clays consist of silica tetrahedra and the octahedra contain magnesia surrounded by oxygen atoms and hydroxyl groups. They are often found mixed with other clay minerals and with calcareous materials. They fall into two groups: the attapulgite-palygorskite group and the sepiolite group.
In attapulgite clays, the magnesia (magnesium ions in octahedral units) are mainly replaced by aluminia ions. The minerals are in the form of laths or bundles of laths, several microns in length, often in a bent and tangled pattern; they are highly absorbent. In paly-gorskite clays fewer magnesium ions are replaced by aluminia in the octahedra, than in attapulgite clays.
These clays contain octahedra with even fewer magnesia ions being replaced by aluminia than in palygorskite clays and the laths are thicker, shorter and denser than in attapulgite clays.
The preparation of clays can include mixing different kinds of clays and adding various kinds of inorganic or organic materials (known as temper). But before these processes occur preparation can also include forms of purification such as physically removing rootlets and stones, sometimes by drying the clay, crushing it and sieving it. Another technique that can be used is levigation, in which the clay is thrown into large tanks. The larger inclusions settle out and the finer particles are separated by remaining in suspension. Peacock (1982: 54, 122) describes a Roman levigation tank which had a 10,000 gallon capacity. Smaller installations were also used for decanting and evaporating clays from levigation tanks. During this process inclusions, like calcite, which could cause problems during the firing (see Section 4.3.10), can be removed.
This occurs prior to the physical techniques of wedging, kneading and foot-treading. These physical techniques of preparation eliminate air pockets and increase homogenisation of moisture and inclusions making the clay more workable. Wedging consists of repeatedly cutting the clay with a wire and pressing it together (see Figure 4.2a). Kneading, folding and pressing the clay are also used to eliminate air pockets (see Figure 4.2b). Foot-treading has the same effect as kneading but is carried out when large quantities of clay are involved (see Figure 4.3). These preparatory techniques ultimately aid in the successful firing of the pot and improve one of the most important physical properties of the clays, plasticity. Plasticity can be defined according to its yield point (the point at which compressive forces start to change the shape of the clay) and extensibility (the amount of deformation a clay can withstand beyond the yield point without cracking).
Figure 4.2a Wedging clay.
Figure 4.2b Kneading clay.
Figure 4.3 Foot-treading clay at the Heracla brick works, northern Syria.
With the adsorption of water, a clay can be shaped under pressure so that when it dries the vessel form is retained. When fired above certain temperatures the property of plasticity is eliminated. As mentioned above ( Section 4.2.3 (II)) clay particles are very small, and a large proportion behave like colloids. The platelet shapes of the particles have large surface areas which lend themselves to the property of plasticity; the adsorbed water acts as a lubricant allowing the clay particles to slide over each other, while at the same time they are difficult to pull apart because the surface tension is sufficiently strong. Clay can be compared to two sheets of ice separated by water which slide past each other. Generally, the larger the surface area of colloid-sized platelets, correlated to the incidence of finer particles such as in smectites, the more plastic the clay; kaolinites have larger particles and tend to be less plastic. Excess water lowers the surface tension and makes the clay soft and weak.
Potters judge the workability of a clay (its plasticity) according to feel: its working range and plastic limits, both of which rely on the addition of water in order to create a workable clay from a dry clay. A clay which is to be wheel-thrown should have relatively wide plastic limits because it must be able to flow, while at the same time it should not dry out excessively. Hand-made pots can be made from stiffer clays with a higher yield point and lower extensibility. Some clays, such as montmorillonites are suitable for both wheel-throwing and hand-building and are described as ‘fat’. There are various ways of increasing the plasticity of a relatively stiff clay: the addition of organic material such as yoghurt, beer, bacteria or starch and acids (e.g. vinegar). Plasticity can be increased if a clay is aged so that bacteria and acids allow the water to reach all clay particles and cause flocculation (the agglomeration of particles in the clay, forming ‘flocs’) (Glick 1936). Another obvious way of increasing plasticity is to mix a relatively fine plastic clay with a stiff clay.
Layered silicates, especially smectite clays, can be easily worked and are therefore plastic because water can easily be adsorbed (and see Section 4.2.3 above). This water is so weakly-bound to the clay minerals that it turns to steam at low temperatures. This kind of water is distinct from the water (and hydroxyl groups) which is chemically-bound to the clay minerals and the water which forms between the lattice layers (Norton and Johnson 1944; Dal and Berden 1965).
Rice (1987: 55–7) describes the complex electrical properties of both the clay minerals and water molecules which result in the electrical ‘attraction’ of water to clays. The clay platelets have electrically charged surface sites, such as in kaolinites, for example, where ions of aluminium (Al3+) and silicon (Si4+) are located along the edges of the platelets and oxygen (O2−) and hydroxyl (OH−) ions on their surfaces; these ions are electronically ‘unsatisfied’. These ‘unsatisfied’ sites bond with hydrogen and oxygen ions from the water, forming an adsorbed water layer around the mineral platelet. The adsorbed ions deriving from the water are organised (‘immobilised’) into a layer which in some senses is analogous to ice so that the layers ‘slide’ past each other (Pollard and Heron 1996: 122) and which clearly provide the important property of plasticity.
Thumb-pinched pottery is produced by inserting the thumbs into a lump of clay. The fingers are used to squeeze and pull the clay until it becomes thin enough to form the wall of the pot. It may be carried out on a flat surface so that the base in turn becomes flat, by pressing the clay down onto the surface. Taller and larger vessels are produced by drawing the clay upwards into the desired shape. If the shape of the pot base is to be conical or rounded, then a support/cradle is necessary; alternatively a mould can be used (see below).
Slab-building is carried out by flattening the clay on a flat surface to create an even thickness, perhaps with a roller, cutting it into slabs and then luting the slabs together with a thick suspension of clay (slip). Angular or cylindrical vessels are often built using this technique.
Coil-building is another commonly-used technique where the clay is formed into a series of rolls/fillets by rolling it on a flat surface. The coils are then built spirally on a pre-prepared base until the desired pot size is achieved. Rings and segments may be added to a pot which has been produced using another technique. As coils are added the clay needs to be squeezed to create a join; the addition of a slip to the surfaces may hide the features which are characteristic of the technique. Xeroradiography, however, will reveal clearly the coils used to build the pot (Rice 1987); any cracking which occurs will also be in characteristically horizontal lines following the arrangement of coils.
Moulded vessels are produced by pressing the wet clay into a pre-prepared fired ceramic or plaster mould. The mould fragment shown in Figure 4.4 was decorated using a stamp. The result of pressing the (leather-hard) pot into the inside of the mould is that the outside of the pot takes up the shape and/or the decoration (see Figure 4.6). Moulds can also be used for producing appliqué (see Figure 4.5). By noting slight imperfections in the mould it is easy to identify the existence of a series of pots made using the same mould. The moulds themselves may be single or in two parts. Buffering materials (between the clay and the mould) are sometimes used so that the decorated clay can be removed from the mould with ease. Examples are fine sand, powdered clay and ash. If the clay is applied to the inside of the mould, it should not be allowed to contract too quickly because this can cause cracking as it is compressed; if applied to the outside this is not a problem because it will separate itself by shrinking. If a two part mould is used the seam line is visible on the pot; there is also a contrast in texture between the smooth moulded side and the rougher hand shaped side of the pot. Good examples of mass-produced decorated moulded wares are Roman Samian wares and early Islamic ‘Abbasid wares.
Figure 4.4 Part of an ‘Abbasid mould for a pot, which itself is stamp-decorated, al-Raqqa, Syria.
Figure 4.5 Roman moulds for making appliqué bulls’ heads and dancers
(photo courtesy of R.J.A. Wilson).
Figure 4.6 Roman lamp mould and the lamp made from it
(photo courtesy of R.J.A. Wilson).
The first wheel-thrown pottery, as opposed to wheel-built, had come into common use by 2250 BC, the middle Bronze Age in Mesopotamia. The use of a wheel introduced a kind of control over pot-shaping which was difficult to achieve using other techniques. The distinction between a wheel-thrown as opposed to a wheel-built pot needs some explanation. Wheel-thrown pots are those which are shaped by using a wheel that can be turned mechanically, usually with the feet (at variable speeds, sometimes to rotate a cylinder below the wheel) leaving the potter’s hands free to shape the pot. Alternatively it can be moved with a stick inserted in a hole in the rim of the wheel or with the hands as used by the Ibibio in Nigeria (Nicklin 1981b). Wheel-built pots on the other hand are shaped on a simple turntable (tournette) and may, as a result, be quite symmetrical – because the pot will be centred on the wheel – but the wheel is used as a viewing surface.
To make a wheel-thrown pot the clay is thrown onto the centre of the face of a flat wheel where it sticks while it is rotated mechanically and, at the same time, the clay lump is formed into the desired shape with the potter’s hands. The clay needs to be quite wet because the act of pressing and forming the clay tends to dry it. Initially the clay is opened as with a thumb-pinched pot (see Figure 4.7), the wheel is rotated rather slowly at this stage as increasing pressure is applied to the outside of the pot with a counter-pressure from the thumbs on the inside. The clay is normally pulled upwards (coning) and shaped with fingers and thumbs into a desired shape (see Figure 4.8) while at the same time the speed at which the wheel spins creates a centrifugal force which needs to be countered. More water may be added periodically in order to keep the clay plastic and shape the pot (Figure 4.9); if too much water is added the clay may become too soft and the wall of the pot may break away from its base, attached to the wheel. At some point the rim needs to be finished off (Figure 4.9). Once completed the pot can be sliced off the wheel with a wire. The clue for identifying a pot manufactured using the wheel-thrown technique is by the presence of fine concentric rings (rills) on the walls of the pot, if they have not been covered or obliterated by later finishing. A combination of moulding and wheel-throwing can be carried out. This is when wet clay is placed over a mould already located centrally on a wheel and the clay is pressed into the mould as the wheel rotates.
Figure 4.7 Thumb-pinching the clay.
Figure 4.8 Coning the clay.
Figure 4.9 Finishing the rim of the pot.
The kick-wheel consists of a small light upper wheel on which the clay is placed and shaped by the potter and a much heavier lower bearing or fly wheel. The weight of the fly wheel can provide so much momentum by being kicked judiciously that it can rotate continuously. The evidence for the earliest use of the potters wheel is meagre: its use has been inferred from the analysis of the layout of painted pottery designs dating to the end of the Halaf period in Mesopotamia (Nissen 1988: 46–7). From some-what later contexts parts of a number of potters’ wheels have been found: a basalt pivot which would have formed part of a bearing was found in a Canaanite (1200–1150 BC) potter’s workshop (Magrill and Middleton 1997: Figure 6a). A fired clay potter’s wheel weighing 44 kg (20 lb) was found in the potter’s quarter at Ur. This had holes near its edge in which a stick would have been inserted in order to turn it (Woolley 1956: 28; Simpson 1997b: Figure 1). In medieval Haarlem, the Netherlands, van der Leeuw (1975) has suggested that a ‘cart wheel’ was used, with a turntable set in the middle.
Even though the potters wheel was introduced from a relatively early time in the development of pottery technology, hand-made pots were still, of course, being made. In spite of a lack of direct archaeological evidence for the kind of potters wheel used in medieval England, Singer et al. (1956: 288) published a thirteenth-century French illustration of a potter turning a potters wheel with a stick. Handmade pottery was, however, still being produced at Ham Green, Bristol and Lyveden, Northamptonshire in the thirteenth century (Cherry 1991: 201).
The paddle and anvil technique involves beating the outside of the vessel with a stick while a stone or clay ‘anvil’ is held at the opposite side. This can compact the clay, thin the walls, smooth or roughen the surfaces, remove the marks left by coil-building and improve bonding between separate segments used to make the pot; the technique has, for example, a long history in the manufacture of Chinese pottery (Wood 1999: 18–19). Scraping the leather-hard vessel surface will also remove imperfections and irregularities. A knife may be used to trim leather-hard pots of any excess, particularly from wheel-thrown and moulded vessels.
By using something like a damp cloth, the outside of a leather-hard clay vessel can be smoothed, alternatively the vessel may be wetted first. Burnishing leather-hard or dry clay may be achieved by rubbing the surface with something smooth, like a pebble or smooth piece of bone. Smooth pebbles were, for example, found in a Canaanite potter’s workshop at Lachish (Tell ed-Duweir) in Palestine (Magrill and Middleton 1997: Figure 1). The process of burnishing actually realigns the fine clay particles as well as compacting the clay surface.
Various tools (often made from bone) can be dragged across the surface of a leather-hard clay vessel in order to decorate it. Combing, stamping, impressing, striating and rouletting are some possible techniques which can lead to an infinite combination of decorative designs. Cord has been impressed into the surfaces of beakers in order to produce the famous Beaker ware dating to about 4500 years ago. Stamping and impressing can be performed with the decorated end of a bone tool (akin to a die), a shell or a host of other possible objects that when repeatedly stamped into the surface repeat the same pattern in a controlled or random pattern. Rouletting is performed with a decorated cylindrical tool which is rolled across the pot surface (Balfet et al. 1983: 101).
The pot surface can also be carved, incised and perforated: all three of these techniques involve the removal of clay. These operations are generally carried out when the clay is leather-hard. Incised designs can, of course, be infinite. The kind of incisions produced, fine incisions, narrow grooves, or gouged lines, will depend on the kind of tool used. Incising leather-hard or wet clay can displace a small amount of clay along the margins which may remain when the pot is fired; incising which has been carried out once the pot has been fired causes chips to be removed from the margins of the lines. Examples of incising are combing and scrafitto; the former produces a series of ‘combed’ lines, the latter involves incising decoration through a slip layer which is then sealed with a layer of glaze and fired.
Carved decoration can either be carried out relatively deeply so as to remove areas of clay and to produce a decorative design or in a more shallow way (plano-relief, champlevé or excising) so that the remaining clay is seen to form the decoration in low relief. The clay may be cut away in chamfered sections so as to create panels which can be stepped.
Appliqué decoration involves the addition of small preformed elements to a pot surface such as fillets, pellets and spikes normally luted with a small amount of slip (see Figure 4.10). Handles, feet and bases may be similarly luted into place and can be both decorative and functional. Barbotine decoration is the application of a thick suspension of clay to the pot surface using a similar technique to that used for cake-decoration whereby the clay is carefully squirted into place on the pot surface creating raised designs.
Figure 4.10 Islamic (11th century) carved pot with appliqué from al-Raqqa, Syria.
Slip decoration is performed with a liquid suspension of fine clay particles which is often of a different colour from the pot being decorated. The slip may provide a smooth surface for subsequent decoration (including underglaze painting), but also a consistent, usually pale, colour providing a decorative contrast. An example of the use of slip to produce some highly decorated pottery in antiquity was that used for the manufacture of Roman red slip decorated terra sigillata (Roberts 1997: 189ff.), which includes Arretine and Gaulish Samian wares. Fine textured illite clays were often used in the manufacture of sigillata and the particles, in that case, may be held in suspension because they are sheet-shaped. In compositional terms they often contain higher alkali (potassium or sodium) levels than the body of the pottery they decorate and as a result tend to fuse at (slightly) lower temperatures. The slip layer on sigillata is as thin as 30 microns. This family of pots is not only famous for the use of a brightly coloured red slip but is often decorated with moulded motifs or scenes.
Slips can be applied by dipping the pot into a slip, pouring the slip onto the surface or wiping it onto the pot surface. If a pot surface is irregular the application of slip will fill in those irregularities, and create a smooth surface. The slipped surface itself may be incised to reveal the underlying clay. As with glazing, the relative degree of contraction of the pot and slip need to be matched so that the slip adheres; some slips may be ‘tempered’ so as to come close to matching the contraction of the pot. If the coefficients of contraction and expansion are very different, the slip can easily flake off or ‘craze’ in the same way that glazes do. An example of the use of a white slip is in the production of Ottoman Iznik ware (see Figures 4.41 and 4.42; Section 4.7.8) in which a proportion of lead-rich ‘frit’ (actually glass fragments) is added to the slip so that its expansion and contraction matches both the pot body containing ‘frit’ and the translucent glaze which covers it.
If the pot is air-dried before the slip is applied it will extract water from the slip which produces a ‘concentrated suspension’ of clay particles on the pot surface (Velde and Druc 1999: 86). A slip can be distinguished from a ‘wash’ by the fact that the slip is fired on, whereas a wash is a pigment or lime-based solution which is applied to a fired pot (Rice 1987: 151).
Glazes have the same material characteristics as many glasses. For a discussion of the chemical and ‘structural’ characteristics of glass see Section 3.1; needless to say glazes are amorphous, can be coloured by adding transition metal ions and can be opacified with various crystalline compounds. As with glasses, the chemical composition of glazes affects their melting temperatures, that is, the temperatures at which they become viscous. This is an important property if one is considering also the rate at which the pottery substrate contracts as it cools. Glaze chemical composition also affects its absorption of light. However, one obvious difference between glazes and glasses is that glazes are made specifically for their attachment to pottery surfaces. This difference in function can lead to glaze compositions which are not found among ancient glasses at all. Not only is the concept of the mechanical attachment of the glaze to the pottery surface important, but so are the ways in which wavelengths of light may be reflected from different glaze colorants and from pot surfaces. The raw materials that were used to make glasses and glazes will partly have been dependant on the niche that glazing and pottery technology occupied in the particular context being considered, bearing in mind the potential social, economic, ecological and ritual features of that production sphere. If both glazed pottery and glass production occurred in the same production zone, the glaze technology may have been viewed as a natural extension of glass technology or vice versa; alternatively conservative elements of production procedures, be they ritual, social or economic, may have prevented sufficient communication between artisans involved to have brought about cross-fertilisation between the uses of raw materials for making glass and glazes. On the other hand, fuel consumption and the raw materials used in the construction of kilns/furnaces are more likely to have borrowed from each other.
When a pot is glazed it typically undergoes two firings. The first, usually at c. 900–1100 °C is called the biscuit (or bisque) firing of the unglazed pot. It is designed to make the pot stronger, allows it to undergo most of its shrinkage and makes its surface more porous in order to promote interaction between glaze and pot surface. The second firing occurs after the pot has been allowed to cool down and the glaze is applied. This is carried out at a variety of temperatures, depending on the glaze and pottery types. A low melting point glaze may be applied to a high-fired pottery body, though it is clearly very difficult to apply a high melting point glaze to a pottery body with a low maturing temperature.
A critical temperature, which provides one way of classifying glazes, is their maturing temperature. This is the temperature at which the glaze is fused to the pot body and, as the pot cools, develops fully its characteristics of strength. It is, of course, dependent on the chemical composition of the glaze – the raw materials that have been mixed in order to create it and the ‘impurities’ that have been introduced. Another way of classifying glazes is by the kinds of pottery on which they are used.
Homogeneity of glaze recipes is equally important to the final colour achieved in glazing. Variations in firing temperature, firing atmosphere and firing cycles can all modify the glaze colour achieved as indeed does the basic glaze composition — whether colouring compounds have been deliberately introduced or not. To be able to coat a pottery body with glass in such a way as to make it stick is quite an achievement. The reason for this is that the expansion coefficient of the clay is normally much lower than that of the glaze. A crazed network of cracks (crazing) can be produced if the glaze contracts more than the body, shivering is when the glaze falls off the body because it contracts less or at a slower rate. The main aim is for the glaze to contract slightly more than the body of the pot so that it fits successfully and compresses the surface of the body slightly. Clearly the relationships between firing technology, and the structural and chemical characteristics of glazes and pottery bodies are both complex and intriguing. The other factors to be taken into account when producing a glaze are the maturing temperature of the glaze (determined by its chemical composition), its surface tension and, related to this, its wetting properties.
Various flaws in the glaze can be created if it is fired at the wrong temperature: if it is too high it may simply run off the pot surface or generate bubbles (though this may be a deliberately produced effect). Under-fired glazes may be dull and do not have a smooth glassy feel. Flaws in glazes can be produced if the fit of the glaze on the pot (i.e. the match of the expansion and contraction of the glaze and the pot as the two are heated up and cooled together) is not achieved.
The classification of glaze chemical compositions (and their ‘melting’ behaviour) can be based on the relative levels of: (1) lead oxide (some of the lowest melting temperature glazes); (2) alkalis like potassium and sodium oxides; (3) alkaline earth metals especially calcium oxide; and (4) aluminium oxide – high levels create glazes with some of the highest melting temperatures.
The earliest pottery glazes date to c. the middle of the second millennium BC in the Middle East (Moorey 1994; Peltenburg 1971). Their basic composition was alkali-lime-silica. This is consistent with glass technology at the time – the principal alkali used was also soda. However, the result of using a soda-rich glaze was that it tended to produce cracks in the glaze as the glaze shrunk onto the pot (i.e. the glaze fit was poor). Indeed Hedges and Moorey (1982) were (at the time) surprised to find that all of the pre-Islamic glazed pottery dating from between 600 BC and AD 600 from Kish and Nineveh that they analysed had been made in the soda-lime tradition.
Freestone (1991) reports finding the same soda-lime technology in ninth-century BC glazes from the Neo-Assyrian sites of Nimrud, Ba’shiqa and Arban. During the eighth and seventh centuries BC there was a peak in the production of polychrome blue, white and yellow glazed pottery in Assyria and Babylonia. These findings were entirely consistent with the analysis of the glazes attached to Neo-Babylonian bricks from the Processional Way in Babylon (Matson 1986).
Although in the Middle East alkali-lime-silica glazes continued to be produced, some of the first lead-rich glazes were introduced by the Chinese in the Warring States period (475–221 BC) on earthenware jars; in this instance they are lead oxide-silica or lead oxide-baria-silica (Wood and Freestone 1995) and contained c. 20% lead oxide. Slightly later, in the Han Dynasty, glazes contained significantly higher lead oxide levels of typically c. 53–60% (Wood 1999: Table 78) which, with the balance of c. 29% to 33% silica and 3.5% to 6.7% aluminia, is close to the eutectic. (A eutectic mixture is one which has the lowest freezing point of all possible combinations of its components.) The Romans, who also used lead glazes and perhaps introduced the technology independently, may have had a (late) influence on the development of Chinese lead glazes, but it seems likely that the Chinese were the first to introduce lead-silica glaze technology. The Roman glazes contained c. 45–60% lead oxide (Tite et al. 1998: 242). These lead-rich glazes continued to be manufactured in the Byzantine and Islamic worlds and sometime later, in the sixteenth century, high lead enamels used for cloisonné on metal vessels produced in China have been found to have a similar composition to the glazes used on fahua wares – the Chinese ceramic versions of cloisonné decoration on metal (Wood et al. 1989). This reminds us that there may have been a range of levels of technical interaction between those involved in vitreous technologies.
A lead-alkali glaze technology was first added to the alkali glaze technology, probably in Iraq, in the eighth century (Mason and Tite 1997). Although the Romans produced lead glazes in various shades of green due to the presence of iron, the use of lead by Islamic potters tended to be associated with the introduction of the earliest tin opacifiers in pottery – a gap of some 900 years since the introduction of tin as an opacifier in glasses (see Section 3.2.5.5 ); tin was also used intermittently by late Roman glass-makers. The lead oxide-soda-lime Islamic tradition continued in use and was employed for sixteenth-and seventeenth-century Iznik glazes (Henderson 1989b; Tite 1989). A ‘western’ equivalent was used in the manufacture of lead-rich glazes of the Italian Renaissance in which potassium oxide replaced soda (Tite 1991).
In the medieval west the familiar olive green colour of so many of the wares is due to the presence of iron in the pottery which interacted with the lead-silica glaze under reducing conditions. In addition to this, yellow and colourless glazes were produced under oxidising conditions (Newell 1995); the use of a colourless glaze would simply provide the pot with a glassy appearance with the orange or reddish clay being visible behind it. The darker green hues found in this medieval pottery were produced by using copper and brass filings. The addition of iron oxide produced a darker brown colour in the glaze under reducing conditions. Because these glazes sometimes contained high (< 60%) lead oxide levels, the potter had to be careful not to reduce the glaze to the extent that the lead was reduced to metal; this would have produced a matt metallic finish.
The glaze would have been applied in the form of a paste by brushing it onto the pot (as powdered litharge or lead sulphide – galena PbS), dipping the pot in the glaze or sprinkling a powder on it. In the eleventh and twelfth centuries glaze appears to have been splashed on (Cherry 1991: 202). Analytical investigations of medieval pottery found at a kiln site of Hanley Swan, Worcestershire, UK has revealed that lead oxide was added in its ‘raw’ state to the body of the pot and a white layer formed as a result of the lead reacting with the body (Hurst and Freestone 1996). However, one result of brushing a powdered lead compound onto the pot is the toxic lead fumes which are produced when the pottery is fired. A way of preventing this is to fuse the lead with glass and then powder the glass, which is then applied to the pottery surface. This material is labelled ‘frit’ by some potters and art historians, but strictly speaking frit is defined as the partly fused raw materials of glass production, so the label is incorrect in this context. Fusing the lead in the glass prevents further lead fumes from being generated, but they may still be produced when the glass is made in the first place. There is however another advantage to applying powdered glass to a pot surface to produce a glaze: the resulting glaze will be homogeneous with fewer flaws.
Another type is salt (NaCl) glaze. This was produced by introducing salt into the kiln at high temperatures. A proportion of the salt decomposes with the alkali (Na) combining with the aluminium and silica in the pottery to produce a soda-rich high-temperature glaze; the chlorine in the salt combines with hydrogen in the water (H2O) molecules in the pot to produce an acidic hydrogen chloride gas (Starkey 1977: 2).
In the Far East the glazes used on stoneware and porcelain tended to mature at high temperatures (up to as high as 1350 °C) having different chemical compositions from the soda-lime and lead-rich glazes used in the west (Wood 1999: 38; see Section 3.6). The principal compositional difference is the inclusion of a high alumina content with relatively low (resulting) levels of calcium oxide, soda and potassium oxide. In some Chinese pottery between c. 1500 BC and AD 900 the alkalis, a mixture of soda (Na2O) and potassium oxide (K2O), although at a maximum level of 5%, was provided by wood ash applied to the surface, which interacted with silica in the pot and produced a mottled colour. Some glazes contain elevated magnesia which increases their durability; this was certainly necessary given the low calcium oxide levels. After about AD 900 in China ‘lime’ glazes are supposed to have supplanted ‘wood ash’ glazes (Wood 1999: 32). However, in compositional terms the two ‘compositional families’ are virtually indistinguishable (Wood 1999: Tables 6 and 7), with calcium oxide levels of c. 16–20% and total alkali levels of between c. 2 and 3% in both. Further scientific research into the recipes used and the production procedures employed is clearly necessary.
Many of the colorants used in glazes were also used in glasses (see Section 3.2.5). Moreover many of the factors which determine the colour of glazes are the same as those which affect glass coloration, the only differences being (a) that some glaze chemical compositions are not found among glasses and (b) the presence of a pottery body. The actual physical explanation for coloration in glasses and glazes, in terms of the chemical environment in which the colorants act is given in Section 3.2.5. The application of painted designs for which pigments are ground up and painted on generally relies on the pigment particles not dissolving in the glazes.
Painted designs can be applied either before or after firing. The pigments used can either be inorganic (made from ground up coloured minerals), or organic. Pigments are often made into a suspension with water, fine clay and a binder, and then painted onto the pot surface. Of the mineral-rich pigments, iron-, manganese- and carbon- (ground charcoal) based colorants can survive the firing process. The iron-rich minerals produce brown, red and black colours; an example of a black colorant is irontitanite (ilmenite). Manganese-rich minerals can be used to produce black or brown colours. The colours of these mineral-rich colorants essentially remain the same throughout firing (Goffer 1980). Such minerals were used to produce paint colours for decorating prehistoric pots in the Middle East and present day Pakistan (Wright 1986). Other mineral-rich colorants are used in glazes (see Section 4.3.5). Organic paints are often applied with a brush after the pot has been fired. These break down and lose their colour in the kiln. An example of such a paint is Maya blue which consists of a suspension of clay mixed with metallic oxides; the organic dye is absorbed onto the clay substrate (José-Yacaman et al. 1996). On the other hand resist painting involves the use of a protective coating of something like wax over the parts of the pots which are not to be painted – and which is removed by heat (Shepard 1976: 206ff.). Slips can also be ‘painted’ on.
There are, however, studies of pottery glazes which underline some of the differences (and similarities) between glass and glaze coloration. Discrete changes in the use of cobalt-rich colorants have been identified in the glazes produced between c. 1430 and 1650 found in Turkey, including Iznik (Henderson 1989b: 67). Both manganese-rich and arsenic-rich cobalt ores were used at different times. The use of colorants not used in glasses can be noted: in Neo-Assyrian ninth-century BC pottery Freestone (1991) found that a black colour was caused by a Mn (Fe) pigment. However other colorants he found are also found in contemporary and earlier glasses: the white (opacifying) colorant used was calcium antimonate, the yellow (opacifying) colorant was lead antimonate and an opaque green glaze was produced by a combination of iron and opacifying crystals of calcium antimonate. In Partho-Sassanian glazes Hedges (1982) found that a blue colour was due to reduced iron, green was due to iron (with or without copper), darker browns due to iron and manganese and a black colour due to iron sulphide which had been produced under reducing conditions.
Kingery (1986) has underlined the importance of painted underglaze pigment technology in the Italian Renaissance, making the point that before the fifteenth century (in the west) most glaze colorants dissolved in the glaze and had blurred boundaries. Although, as mentioned above, the use of crystalline opaque colorants in glazes was not new, the definition of the areas to be decorated was. In around 1557 Cipriano Piccolpasso (1980) published The Three Books of the Potter’s Art, an invaluable source of information about the kinds of raw materials used to make the glazes and glaze colorants for Renaissance pottery. In it there is a description of the preparation of two pigments used: lead antimonate (yellow) and tin oxide (white). Neither of these opacifiers were new to vitreous technology, the former being used first in the mid second millennium BC and the latter used first in the second century BC (Henderson 1985). However in this context the opaque crystals could be applied as a paint in areas defined by sharp lines. Other opaque colours were produced by mixing copper with lead antimonate (green) and iron with lead antimonate (orange), again both effects which had already been achieved in ancient glass coloration. A new pigment, called bole red, a ferruginous material, was used both in the coloration of Iznik underglaze decoration (see Section 4.7.9) and in pottery made in Renaissance Italy, appearing for the first time in Iznik pottery in ‘Rhodian’ style pots of about 1580. It was found to consist of finely ground particles of an iron-rich pigment mixed with fine silica particles, producing a red colour under the correct furnace conditions (Henderson 1989b: 68, Table 2). The opaque yellow/orange colour sometimes observed in ‘bole red’ suggests that furnace conditions played an important part in determining the final colour achieved.
Another pigment used by the potters of Iznik for creating (and defining) an outline was a mixture of iron, nickel, copper and cobalt (Henderson 1989b), quite possibly ground-up smalt, a pigment used for colouring medieval glass and also used by artists. A very dark green coloured pigment also used for outlining applied decoration was chromite (Tite 1989). This continued to be used for colouring Iznik glazes into the mid seventeenth century, though at this time the glaze technology had declined (Henderson 1989b: 68). It is not until the nineteenth century that chromium was introduced into the repertoire of true glaze colorants.
Various factors affect the ways in which clay dries: the amount of water in the clay, humidity in the immediate environment, other characteristics of clay particles (their size, flocculated or deflocculated and their orientation), the temperature at which the drying occurs, any variations in temperature, and the air currents. The ways in which pottery dries will now be described, especially what happens to the water in relation to the physical structure, and the manner in which this affects the strength of the pot.
When pottery dries in air the adsorbed water which surrounds the clay platelets and which provided lubrication when the pot was being shaped, starts to evaporate. In so doing the clay platelets draw together until they are no longer separated, increasing the density of the clay and small cracks may develop (Kingery and Francl 1954). The clay is now ‘leather-hard’ and is no longer plastic, so it can be easily handled. Some of the water which was added to make the clay plastic is held in pores in the clay and this takes longer to evaporate; the finer the particle size in the clay the finer the pores and the longer it takes for this water to evaporate. Pore water is replaced by air and when this water evaporates no further contraction occurs (i.e. the bulk volume remains the same). Evaporation through pores starts at the surface with water being continuously drawn out and evaporating as a moisture gradient comes into being. One potential result of fine clays drying too fast is that water evaporates from the surface of the pot faster than it can be supplied from the interior; the result is that the surface contracts faster than the interior sometimes producing compression of the surface which can lead to cracking. This process can be controlled if the clay is allowed to dry slowly. As the clay dries the moisture gradient may also draw out salts and organic particles which will be deposited on the surface as a scum which can affect the colour, density or hardness of the pot (Brownell 1949). The occurrence of non-clay inclusions such as sand or shell particles produces a more open structure, even if there are fewer pores (Mehren et al. 1981). These particles have no water films around them, so less water is held per unit volume than found in fine clays, but because shrinkage of the clay is relatively low and movement of water through to the surface is easier, drying is easier. What is of particular significance is that when inclusions which have no water films around them are added to clays, shrinkage of the clay is minimised and so therefore is the potential of the pot warping and cracking.
Water is also held in a layer one molecule thick on the clay surface (surface adsorbed water) even when the clay may appear to be ‘dry’ (Rice 1987: 65). The water held in the mineral lattice of the clay (see Section 4.3.2) is lost when the pottery is fired.
The distribution of water in the body of the pot, and the orientation of the clay particles within it, are both factors which affect the way in which the pot shrinks as it dries. When a pot is produced by wheel-throwing the continuous addition of water to the outside leads to differential water content; when the pot dries the outside will shrink more than the interior which can lead to cracking. When clay particles are in a random arrangement the clay is more likely to crack or warp. However, when a pot is shaped by coil-building or wheel-throwing the clay particles can line up into patterns known as preferred orientation. This is when the particles are dragged into particular orientations by stroking the clay repeatedly while shaping the pot; even when a pot is moulded the absorption of water by the mould can realign the clay mineral particles into particular orientations. This helps to reduce shrinkage and cracking. Anisotropic shrinkage is another cause of cracking. As the pottery dries and contracts the pot may contract at different rates because during the formation of the pot different proportions of water and/or different alignments of clay particles are concentrated in different parts of the pot.
The term green strength when applied to pottery refers to its resistance to cracking and warping as it dries in the unfired state (Ryan 1965). The parameters which may add to this strength are particle size (the finer the particles the greater the strength), the presence of sodium ions (Na+) and organic materials (like gums and milk solids) and a deflocculated state (where the particles do not form agglomerations [flocs]). Green strength is at its greatest when clay is completely dry. However, a balance needs to be struck between shrinkage and the attainment of green strength: deflocculated clays have lower shrinkage than flocculated clays as well as greater green strength. Fine clays, which have greatest green strength, also shrink more than clays containing coarser particles. A compromise is to add coarser particles so as to promote evaporation of water and the movement of water from the interior to the exterior; the result is to reduce the risk of cracking. Obviously in practice different potters get to know how different clays behave and modify the preparation procedures accordingly.
The texture of clay can be assessed simply by rolling a small quantity between the fingers; if it feels smooth or ‘slimy’ then it is usually a fine clay, whereas a stiff clay may feel coarse and gritty. The clay can also be nibbled in order to test it. Particle-size fractions of clays can be measured using the suspension or sieving method. The texture is mainly determined by the proportion of coarse grains – the clastics – as well as their size and shape. Clastics are not to be confused with the material that potters add to clays deliberately known as temper or grog (see Section 4.3.8). Clays can contain a wide range of inclusions ranging from pebbles and gravels to the more frequent sand and silt. Sand often contains more than just quartz: micas, feldspars and ferric minerals commonly occur and contribute to the impurities which end up in man-made glass (see Section 3.2.2). Sedimentary geologists distinguish between two sedimentary size ranges: ‘mud’ (containing particle sizes of less than 0.06 mm) and ‘muck’ (sediments associated with peat deposits and other wet environments containing 50% or more fine organic matter).
In the past, potters may have deliberately exploited clays which contained angular inclusions in a range of sizes, since these would bond and dry well, and confer strength on the final fired pot. Modern potters add materials to commercially produced refined clays, producing strength and plasticity in triaxial bodies (Norton 1970: Figures 12.2, 12.3 and 12.4). The three components are: clay(s), a filler (the same as a temper providing limited shrinkage, reduced cracking and reduced drying time) and a flux – normally an alkaline mineral which is a ‘cementing’ component.
The smallest size range of clay inclusions includes colloids which are particles of 0.001 mm (1 μm) or less in diameter suspended in another medium (for example, clay minerals in water) which repel each other because of their electrical charges (see Section 4.2.3(II)). Colloids have the property of being able to adsorb ions (Rice 1987: 76); the finer the clay the greater the potential for this. Ions fill the electronically ‘unsatisfied’ sites on particle surfaces and in so doing affect the arrangement of clay particles which may cause either clustering of particles (flocculation) or the opposite (deflocculation) (Johnson and Norton 1941). This therefore has an effect on the working properties of the clays. In deflocculated clays, where the particles repel each other, the particles are stable suspensions that resist settling, but when the clay dries or the clay particles settle the clay has a low shrinkage and a high green strength; deflocculated clays are therefore ideal to make slips from (see Section 4.3.4). The presence of organic colloids changes the acidity of clays and this (the pH balance) can affect its plasticity and flocculation/deflocculation. In general their presence increases the clay’s plasticity. Most clays contain a small amount of organic matter in any case, with sedimentary clays containing as much as 10%.
Clay is composed of a range of constituents which determine its properties as it is worked, dried and as the kiln or furnace temperature rises. Materials can also be added to clays in order to improve these properties, for example the all-important property of thermal shock resistance. While residues of parent rocks in primary clays and the introduction of particulate material during re-bedding of clays can provide inclusions, there are also some materials which are relatively easily identified as having been added intentionally as temper. However, because of the range of possible origins of the inclusions found in pottery – and in spite of sometimes being able to identify with certainty the inclusions that have been added intentionally – identifying temper has its own problems (Rice 1987: 408). The identity, size, shape and relative amount of the inclusions help to characterise the clays.
Over and above the straight ‘deterministic’ approach to the study of the choice of temper – that is the assumption that raw materials (temper) were selected because they determined the appropriate physical property of the pot when fired and when used (O’Brian et al. 1994: 261) – there are the important considerations of ceramic ecology and cultural considerations. Professor Fred Matson, one of the ‘founding fathers’ of ceramic investigations, has pointed to the availability of ceramic raw materials for pottery production as clearly being an important parameter in determining what raw materials were used (Matson 1965) – a branch of ceramic investigations that has become known as ‘ceramic ecology’ (Kolb 1989). Just as important, and in some cases, even more important, are the cultural choices that a potter may make as to what kind of raw materials are used. Craft practice in this case may be tied to the manufacture of specific technological styles, which themselves can drive production as part of the identity of the craft group involved (Wright 1986). These cultural choices in using particular styles may just as much be tied to traditions of place and time of production as other factors, as has been demonstrated by a study of changing fabric types of prehistoric pottery in the East Midlands in England (Allen 1991). As Sterner (1989: 458) has pointed out, the re-use of pre-fired pot (‘grog’) may have been a way of transmitting the life of an older pot and associated traditions from one generation to another, as found among the Sirak Bulahay of North Cameroon. Akin to this practice is the discovery of a recycled Bronze Age pottery fragment in a Deverel-Rimbury vessel at the cremation cemetery of Pasture Lodge, Long Bennington, Lincolnshire, England (Allen 1991) and in Bronze Age accessory cups (Allen, forthcoming).
Temper can either be inorganic or organic in nature: organic temper includes dung, straw and chaff, but perhaps the commonest kind is inorganic. Among inorganic temper (also described as aplastic inclusions) there are many possibilities: shell, sponge spicules, (quartz) sand, sandstone, limestone, basalt and volcanic ash. Small fragments of pot are also added – known as grog. The material which can be identified with some confidence as having been added deliberately is grog. On the other hand, some volcanic ash and vegetable matter burns away completely during firing, leaving only their casts (spaces) in the pottery. The materials which can be both added as temper and which occur naturally include quartz, calcite, sponge spicules, shell and mica. The distinction between the use of temper and the natural occurrence of the same material in the clays used can however often be made by determining the size, shape (angularity) and amount found – there is generally far more shell, for example, if added as temper (up to 30%). Magetti (1982) has gone further using size as a criterion alone in an attempt to distinguish between deliberately added and naturally occurring inclusions.
Shell or calcite has been used quite extensively as temper (Rye 1976; Arnold 1985). In low-fired clays while calcite expands at about the same rate as the clay, quartz expands at a greater rate. With the close match in expansion one might therefore expect calcite to have been used in preference to quartz as a temper (Rice 1987: 410), but Woods (1986), for example, has found that in the production of cooking pots, this has not always been the case. Quartz is often considered to have been added as a temper to clay if it is angular (because it has been prepared by crushing it) rather than having a rounded profile due to natural wear during transport and deposition in the environment. However, angular quartz also occurs naturally in primary or in sedimentary clays.
Three important factors will affect the changes which take place in a pot as the temperature rises: (1) the length of firing; (2) the temperature of firing; (3) the atmosphere in which the firing and cooling take place. All three of these factors must be considered together; to consider them separately is a somewhat artificial distinction since they can be interrelated and changing one factor can have an effect on the others. For example, mineralogical and chemical changes in the pottery may occur at lower temperatures if the atmosphere is reducing (oxygen-deficient) rather than oxidising.
(1) The length of the firing is very important and it can be broken into three stages: the period when the temperature is being raised, the period during which it is held at the maximum temperature and the period during which the temperature falls. Obviously fuel is used during the first and second periods but it is not likely to be used during the third. A critical expression of the conditions in which various physical and chemical changes can occur in the clay is known as the ‘work heat’. It is the ‘effect’ (however that is measured) of a given amount of heat on the ceramic in a given amount of time. It is in fact difficult to divorce time and temperature of firing.
(2) The temperature of firing as already described can change during different parts of the firing cycle; defining the length of time that the maximum temperature was held at is certainly important. In the past, potters could be described as being at the mercy of the burning characteristics of the fuels used: some will burn faster and give greater heat than others. Having said this, inevitably the burning characteristics of fuels used would ultimately have been experimented with and used to best effect. The determination of the maximum temperature of firing can be achieved using a number of analytical techniques such as differential thermal analysis (Grim 1968). Of course different clays can withstand different firing temperatures before they vitrify and lose their structural integrity. Indeed different wares can be distinguished according to the temperatures at which they have been fired, such as earthenware, stoneware and porcelain. In firing stoneware and porcelain the combination of the high temperatures and raw materials used brought about a degree of vitrification in the body of the pot.
A further point to bear in mind is that the relative position of the pot in the kiln will determine the temperature that it is subjected to: it may vary by as much as 150 °C (Simpson 1997b: 52). Terracottas which are ‘normally’ fired in open firings are heated to below 1000 °C; earthenwares between 900 °C and 1200 °C and stonewares, at between 1200 °C and 1350 °C. Porcelains, being most refractory, with the highest aluminia content, will fire as high as 1400 °C. Bonfire firings are generally quicker than kilns and a deal less controlled, with their maximum temperatures being held briefly. Although lower temperatures were achieved in bonfires, this doesn’t necessarily mean that all low temperature firing was carried out in them (see Section 4.3.11).
Today potters use clay pyrometric cones of known compositions which bend at specific temperatures. By watching them during the firing cycle it is possible to hold the kiln at the appropriate temperature.
(3) The atmosphere of firing is important because it can affect the colour and hardness of the pot, and, in addition, its shrinkage and porosity. In discussing, or attempting to reconstruct, kiln technology it is important to know what the firing atmospheres were during the period when the kiln was held at its maximum temperature, but also during the cooling phase. The atmosphere may be affected by the differential air flows at different times, the use of different kinds of fuel at different times (such as different species of wood – see Section 5.6.2) and a covering of ash over the pots. The kind of firing set-up used is obviously important in determining the firing atmosphere achieved. The firing atmosphere is defined as the balance of gases (such as oxygen, carbon monoxide and carbon dioxide) during the firing procedure. An oxidising atmosphere is one in which oxygen predominates, whereas a reducing atmosphere is oxygen-deficient (i.e. carbon dioxide and carbon monoxide predominate) and is often smoky. A kiln in which air is able to circulate freely will normally be an oxidising atmosphere (though will obviously contain the other gaseous components of air, such as nitrogen). Sulphur-rich fuel combustion may produce sulphur dioxide gas and water vapour; the pottery itself may produce water vapour and the sulphur-rich and carbon-rich components of the clay minerals in the pots being fired when oxidised during firing may also produce their respective gases – SO2, CO2 and CO. When open firing was used, such as in bonfire firings (see Section 4.3.11.1 ), the control over air supply and the rate of fuel consumption would have been more difficult than in a pottery kiln. The net result (e.g. pottery colour) would have been less predictable. In open firings the atmosphere is often a compromise between oxidising and reducing, being neither completely oxidising nor reducing; to produce a completely reduced black colour caused by the oxidised (ferric) iron being reduced to its reduced form (ferrous), prolonged firing in a reducing atmosphere above 825 °C is necessary (Sheppard 1976: 219). On the other hand, in a kiln, control over the atmosphere was easier: a completely oxidising atmosphere was possible if a free flow of air was attained; to achieve a completely reducing atmosphere was more difficult – the fuel needed to be ‘smothered’.
Irrespective of the firing technology used, when a dry pot is heated, the remaining water held in the clay is gradually lost and at the same time the clay mineral structure and its chemical characteristics will be altered. During these processes the clay becomes harder.
Most weight loss on firing is due to the loss of surface adsorbed water and pore water from the clay (see Section 4.3.2 above). Cracking may occur if this process occurs too fast. The chemically combined water is driven off at higher temperatures, the process being dependent on the kinds of clays involved and therefore the ways in which water is held within the lattice structures (as hydroxyl groups). The effect of mixing clays appears to make the loss of hydroxyl groups more gradual and reduces the temperature at which it occurs. The volatilisation of organic components as gases (carbon monoxide and carbon dioxide) also contributes to weight loss; a high proportion of organics in the clay will lead to a correspondingly relatively high degree of shrinkage. Incomplete oxidation of organic-rich clay components may lead to ceramics with a dark core when seen in a broken section.
Other causes of contraction include the breakdown of inorganic inclusions: for example, sulphates, sulphides and carbonates, such as calcite (CaCO3), dolomite (CaMg(CO3)2) pyrite (FeS) or gypsum (CaSO4.2H2O), which, when heated, give rise to gases such as sulphur dioxide and carbon dioxide. Chlorides can react with iron in the clay producing iron chloride. As these changes occur above temperatures of 500 °C, the increasing density of the clay is accompanied by decreasing porosity (the pores contract and close).
At still higher temperatures new minerals form and vitrification within the clay starts to occur, eventually producing a glassy phase. New minerals mainly form at temperatures above 900 °C, so are not normally observed in pottery fired in bonfires. Above 900 °C clay minerals lose all water, they break down and form new silicates. For example in kaolins the metakaolin formed at c. 500 °C breaks down into spinel, an aluminium-rich alteration product, and silica, both of which cause increased shrinkage (Grim 1968). It is normally claimed that spinel forms needles of mullite (3Al2O3.2SiO2) between 1050 °C and 1275 °C and at even higher temperatures between around 1275 °C and 1460 °C, yet more shrinkage occurs as more high-temperature minerals are formed such as cristobalite. However, clay impurities, such as potassium oxide and calcium oxide reduce the temperature at which these minerals form (Johnson et al. 1982). A.C. Dunham (1992) has pointed out that the soaking time is an important factor in the development of high-temperature crystallites in the ceramic: mullite spinel assemblages form at 950 °C after only two hours soaking at that temperature; at higher temperatures the amounts of mullite spinel assemblages increase. So the combination of factors which affects the formation of minerals in a ceramic as the temperature is raised is not straightforward.
These changes in ancient ceramics can be investigated using two principal techniques: differential thermal analysis and X-ray diffraction.
The mineralogical transitions, which have mainly been noted as a result of work with kaolinites, obviously also apply in general to other clay types, such as smectites. The transitions occur at different temperatures in smectites because the clays have a different mineralogical/chemical composition (see Section 4.2.3); their different compositions have the overall effect of lowering the temperatures at which ‘high-temperature’ minerals form — smectites are also less refractory. Montmorillonites retain their lattice structures until c. 800–900 °C. In excess of these temperatures the pathways of mineral development differ according to their compositions: in those with Al3+ ionic substitutions, spinel forms; in those which are generally low in iron, quartz develops instead. As with kaolinites the presence of alkalis (lithium, sodium and potassium) reduces the temperature at which minerals develop and the maximum temperature up to which they survive.
Another group of changes that can occur in pottery as the temperature is raised is what happens to inclusions that may have been added by the potter (as a temper – see Section 4.3.8) or which occur in the clay naturally. The main effect that these inclusions have on the pot, as a result of modifying the microstructure of the clay, is on its expansion and contraction during firing. Finely divided particles of reduced iron, mica and lead minerals can all act as low temperature fluxes. When three types of commonly occurring inclusions – quartz, feldspar and calcium – are heated to sufficiently high temperatures they have specific roles in initiating and developing the partial fusion or fusion of grains. Since this process can have an important effect on the strength, thermal shock resistance, porosity, colour and hardness of pottery, a summary will be provided here.
This fusion process can be defined both as sintering and vitrification. Sintering is defined as a process when particle surfaces (in this case clay particles) begin to fuse or stick to others – either as a solid state or liquid phase reaction (i.e. although the material does not melt completely it forms a coherent mass). In the former a ‘neck’ is formed from the surface diffusion of atoms between the particles – but no liquid phase or melting is involved (Kingery et al. 1976: 469–79). The rate at which solid state sintering occurs is approximately inversely proportional to the particle size involved. Sintering usually leads to a reduction in mass with pores becoming more spherical or being eliminated. In liquid phase sintering, as the name suggests, a liquid phase is involved because components with lower melting temperatures (such as fluxes like alkaline feldspars) melt. More of the solids melt (are ‘fluxed’) as sintering proceeds, and the particles are drawn together as the pores get smaller. This shrinkage increases the density of the pot. Vitrification is an extension of this process as heating continues leading to a glassy phase (Kingery et al. 1976: 490), the elimination of pores occurs and the mass increases even more.
The classic wasters from kiln sites are those where the pots have been heated to such high temperatures that they have become glassy (vitrified), have slumped and in so doing have lost their original shape. With fine-textured ceramics sintering and vitrification occur at the lowest temperatures – an important property when it comes to using slips.
Quartz is a mineral which contains the highest proportion of silica – the quartz sand of glass-makers. Silica is also found in flint, chert, jasper, agate and chalcedony (see Chapter 6). Sandstone, a sedimentary rock and quartzite, a metamorphic rock, both contain a high proportion of quartz. Occasionally the silica-rich skeletal remains of single-celled marine organisms, diatoms, and of sponges occur in pottery. Another form of silica which is found in pottery is organic: phytoliths are present in trees and grasses and remain in pottery once they are burnt. Quartz has been studied in particular in relation to the effect its presence has on the strength, porosity and shrinkage of pottery. A recent study of the mechanical performance of quartz-tempered ceramics (Kilikoglou et al. 1998), mainly with reference to Aegean examples, focused on energy dissipation during fracture and, refreshingly, the discussion covered some of the contextual and cultural considerations for the choice of raw materials. Vekinis and Kilikoglou (1998) have shown that a combination of the Hertzian point loading test and the abrasion wear resistance test on relatively small samples can provide a means of characterising earthenware ceramics which contain quartz and other inclusions. Toughness, an important characteristic of ceramics if determined by the Young’s test, and hardness measurements, can both be related to the results of the Hertzian point loading test.
The presence of quartz in pottery can reduce firing shrinkage, but it can also reduce the strength of the pot when fired. In order to be of benefit it needs to be present as small particles, or in small amounts.
Although quartz is a mineral, and has a high melting point of 1710 °C, when heated it undergoes three kinds of changes in its atomic structure (inversions) which can effect the ceramic structure and its properties. The first occurs at 573 °C when there is a change from α- to ß-quartz accompanied by a volume increase of 2%. Around this temperature the removal of a large amount of water from the clay occurs so it counterbalances any volume increase. The second change occurs at 867 °C, from ß-quartz to tridymite and the third, at 1250 °C, from tridymite to cristobalite. The formation of tridymite and cristobalite occur very slowly, because major rearrangements of silica-oxygen tetrahedra occur; their formation is accelerated by the presence of fluxes. The occurrence of tridymite and cristobalite in pottery is therefore dependent on the temperature being held for long enough during the kiln firing cycle for them to form. Open firing may produce the change from a- to ß-quartz, but may not even promote the formation of tridymite. Crystals which are incompletely dissolved in the glass which forms at high temperatures have a large expansion coefficient which, on cooling and contracting, may cause stresses resulting in cracks in large crystals and micro-cracks in the pottery. Between 200 °C and 270 °C any β cristobalite which formed at 1250 °C reverts to a cristobalite resulting in a volume contraction of 2%.
Feldspars occur in rocks such as granites and lavas. As mentioned above ( Section 4.2.2) they are the most common minerals in the earth’s crust forming 39% of rock-forming minerals. They contain a combination of silica and aluminia and the balance of their compositions leads to a corresponding range of feldspar compositional types. They are the primary parent material for clay minerals and can be present at low levels in clays due to incomplete weathering, imparting some of the properties to the clay described below. In commercial ceramic production clay, (alkaline) feldspars and quartz are mixed.
The three main categories of feldspars which form a continuum of aluminosilicate minerals are characterised by substitutions of potassium (K+), sodium (Na+) and calcium (Ca2+) ions. The proportion of these ions in the different feldspar minerals varies, the result being that potassium oxide feldspars contain an average of 64.7% silica and the calcium oxide feldspars 42.8% silica (Cardew 1969: 44).
Examples of potassium feldspars are orthoclase, microcline and sanidine which have very similar chemical compositions; these are often found in intrusive rocks like granites. Plagioclase feldspars (see Section 4.2.2) can either be sodic or calcic, that is the end member of the sodic feldspar is sodium (Na) and that of the calcic feldspar is calcium (Ca). An example of a sodic feldspar is albite and of a calcic feldspar, anorthite. The continuum of aluminosilicates that these form a part of reminds us that there are plagioclase feldspars with intermediate compositions. Potassium feldspars and sodic feldspars are found in granites and other metamorphic rocks; calcic feldspars, on the other hand, are found in lavas.
The alkaline ions (Na+ and K+) which characterise some feldspars act as fluxes during pot firing, reducing the temperature at which sintering and melting will occur, and they form a thick viscous liquid; this is also helped by their small particle size. Potassium oxide feldspars melt at c. 1150 °C and soda feldspars at c. 1118 °C. The alkaline earth ion (Ca+) in calcium oxide feldspars melts at the much higher temperature of c. 1550 °C (due to the relatively high aluminia levels) – though this is reduced if other feldspars are present.
When calcium oxide or calcium occur in clays naturally, the clay is known as calcareous or marly. The calcium is present in the form of calcium carbonate (CaCO3) as limestone, shell (perhaps as aragonite) and calcite. Calcium is sometimes added deliberately to clays in the form of animal bone ash. The use of bone ash was probably patented by Thomas Frye, a proprietor of the Bow, London Porcelain Factory, leading eventually to the production of bone china (Freestone 1999: 15).
When calcium carbonate is heated it decomposes into calcium oxide and carbon dioxide. The temperature at which this occurs in pottery varies according to the oxidising-reducing atmosphere, the grain size of the crystals and the heating regime, but normally it occurs between c. 700 °C and 900 °C (Velde and Druc 1999: 103). If this occurs above c. 1000 °C, as is the case in many kilns, the calcium forms part of the liquid glassy phase and it can form calcium silicate minerals like wollastonite (CaSiO3) or calcium ferrosilicates. Tite and Maniatis (1975) found that the melting of calcium compounds is accelerated in a reducing atmosphere. Below 1000 °C a potential problem occurs. Calcium oxide absorbs atmospheric water (it is hygroscopic) and eventually reacts with the water to form quicklime (Ca[OH]2). In so doing heat is released, the volume of the pottery increases and this causes the pot to crack and to spall; the strength of the pot is therefore reduced. This effect is particularly marked in clays which contain large lime particles – if they are fine it is less of a problem, and also if the pot is fired in a reducing atmosphere. If the calcium is present in the form of aragonite it will be transformed into calcite at temperatures between 400 and 500 °C (Deer, Howie and Zussman 1962).
‘Open’ pottery firing is also known as bonfire firing and ‘clamp’ firing (see Figure 4.11, 1). Although some authors regard ‘open’ firing to have also occurred in a pit (see Figure 4.11, 2) presumably because there was no superstructure, the definition of ‘open firing’ here only includes bonfire and ‘clamp’ firings. It has recently been pointed out (Gibson and Wood 1997: 58) that there are no ethnographic examples or reliable archaeological evidence for ‘clamp’ kilns. ‘Firing installations’ in which partially-fired vessels are buried in a pit under a mound of soil or turves and which would therefore have reached temperatures of c. 900 °C will not be considered further.
Figure 4.11 A range of pottery firing installations: 1 ‘open firing’; 2 firing pit; 3 through draft; 4 up-draught; 5, 6 and 7 examples of three Roman up-draught kilns (1–4 reconstructed sections); 5–7 cut-away sections showing support structures).
Open firings tend to be short lived and achieve only relatively low temperatures of c. 800–900 °C, with a minimum temperature of c. 600 °C. Cardew (1969: 11) claims that the chemical and physical structures of clays demand a minimum temperature of 550 °C in order to make serviceable pottery, but obviously this varies according to the clays used. By using a pyrometer to measure temperatures in kilns, there is still a somewhat open question as to whether the temperatures measured are those achieved by the pots or the atmosphere surrounding them (Nicklin 1981a). Open firing probably produced some of the earliest pottery – the thirteenth-millennium BC Japanese Jomon pottery (Harris 1997: 20), the tenth-millennium BC Middle Nile pottery of Egypt (Welsby 1997: 27) and the Proto-Hassuna wares of the late seventh millennium BC found in small villages in Mesopotamia (Simpson 1997a: 38).
The pottery to be fired is placed on a bed of fuel and more fuel is placed on top of it. The whole is ignited, and once the firing is over the pots are removed (see Figure 4.11, 1). Ethnographic examples show that individual firing events can include hundreds of pots, such as those assembled by potters in Gogunda, Rajasthan (Rice 1987: Figure 5.17). However, open firing is least easy to control and must have been one reason why pottery kilns were introduced. Evidently if the level of control over the firing is deemed to be sufficient and the products are acceptable in an open firing then there are none of the problems of servicing a kiln. However, open firing is subject to the widest range of problems because it is not contained. As the fuel burns it tends to shift which, in turn, may cause the pottery to shift, raising the possibility of damage. The pottery can be marked by direct contact with the fire as well as by rapid temperature changes which can cause it to crack or to become dented. The extent to which pottery may be overfired or underfired can be related to the way in which it is stacked, and this, in turn, can be linked to an uneven supply of heat, sometimes caused by draughts. As part of the lack of control over the firing process, a lot of heat can be lost to the atmosphere through radiation and convection. In general, non-kiln firings tend to be short, the rate of heating usually not controlled and rather fast, and the time during which the maximum temperature is sustained (the soaking period) tends to be brief before cooling begins (Shepard 1976: 87, 89). One of the most striking characteristics of open firings is the very rapid initial rise in temperature to c. 900 °C – which is attained after about twenty minutes, usually just after the fuel which covers the firing has been consumed. After this the temperature declines rapidly to begin with to about 500–600 °C, and then more slowly. Given the variations of the conditions within an ‘open’ firing, different parts heat up and cool down at different rates, so these figures are generalisations. For these reasons, quite apart from the variations in gaseous atmosphere which could change the glaze colour (see Section 4.3.5), and because of possible physical damage to the outside of the pot, glazed wares are not generally fired in the ‘open’.
Having said this, the technique of ‘open’ firing has been used successfully for millennia to fire cooking and storage pots. The thermal shock referred to above can be avoided by pre-heating the pottery. In the context of a bonfire if the fuel is heated from above then heat is transmitted downwards, although initially a lot is lost to the atmosphere. In any high-temperature process the density of the wood (if used) and therefore its calorific value has a direct effect on the amount of heat produced and the rate at which it burns; experience and experiment would eventually have led to a reduction in the number of unsuccessful firings (see Section 4.3.11.2 ). Shepard (1976: Figure 4) has shown that a temperature of c. 900 °C can be achieved using dung, coal and juniper wood as fuel.
In order to reach higher temperatures than those achieved in a pit the fuel supply was carefully controlled in the context of a carefully constructed kiln. The idea was to enclose the pottery to be fired in a chamber. As for the furnaces used in glass production ( Section 3.4), pottery kilns were constructed out of bricks which could withstand the temperatures and the consequent expansion and contraction when being heated up and cooled. Kilns could also be built out of stone, as in Pereuela, Spain (Peacock 1982: 21).
There are three basic kiln types to be considered: pit kilns, updraft kilns and downdraft kilns (see Figures 4.11, 4.12 and 4.14).
Figure 4.12 A medieval updraft kiln with two flues
© The British Museum.
Figure 4.13 Tripod attached to the base of a glazed pot.
Figure 4.14 A plan of the remains of a downdraft kiln, Five Dynasties (907–60 AD), Yaozhou, China.
Pit kilns are, as the name suggests, elaborations of pitfired installations, but they do not have the superstructure which distinguishes them from the more familiar updraft kilns (see Figure 4.11, 2). In a pit kiln the fuel is placed below and above the pottery and not clearly separated as in updraft kilns. The most prominent difference between pit kilns and firing pits is that in pit kilns (not illustrated) the firing is carried out within a pit which is enclosed by mud or brick walls on three or four sides. The effect of the enclosure walls is to allow the kiln to reach higher temperatures and for longer, than in a firing pit (Shepard 1976: Figure 4). If dug into a slope the draught can be channelled into the pit kiln. Postgate and Moon (1982: 127) report an early Dynastic (early third millennium BC) example from Abu Salabikh, Iraq, lined with bricks which was probably fired with dung cakes or straw. Postgate and Moon (1982: 109) also describe (unlined) pit kilns, each with individual flues from Abu Salabikh dating to the Uruk (fourth millennium BC) period. Because of the proximity of the fuel to the pottery, there would be a risk of disturbing the glaze so unglazed pottery is generally fired in them.
The most familiar kind of kiln is the two-chambered updraft kiln (see Figures 4.11 and 4.12). This is generally a cylindrical construction, sometimes with a domed roof, which is divided into two chambers. The lower chamber contains the fuel; the chambers are divided by a floor in which flues are constructed and through which the heat rises to fire the pottery. The pottery is carefully stacked on the floor of the kiln. The reason why they are called updraft kilns is fairly clear: the draught rises from the fuel into the upper chamber of the kiln. Fresh fuel is either fed directly into the firing chamber or through openings in the side of the kiln. Some of the earliest updraft kilns, with circular plans, were developed in Mesopotamia c. 6000 BC (Simpson 1997a: Figure 2). Those found in agricultural villages such as Yarim Tepe I (Oates and Oates 1976: 42) were used to fire decorated globular jars and open bowls.
Earlier versions of the updraft kilns were not necessarily roofed over but open-topped with broken sherds providing the cover, but which, nevertheless, acted as a ‘chimney’. Apart from the heat which was needed to fire the pots, the other important feature of these kilns was the gases produced by the burning fuel. Because the kilns essentially have open tops a lot of the heat is wasted because it escapes through the top. Another disadvantage is that hot spots can be created, depending on how the pots are stacked in the kiln, although this is almost inevitable in any kiln form. The vessels lying at the bottom of the load receive the greatest heat and there is a gradient of heat, falling off towards the top. When carbon-rich materials in the fuel are oxidised they can produce carbon monoxide and carbon dioxide gases; it is the balance between these gases and oxygen in the kiln which is critical in determining the colour of the (glazed or unglazed) pottery produced (see Section 4.3.5). An advantage of the ‘open’ topped kiln was that the pottery could be loaded from the top. The top of the kiln could have been constructed from pot sherds piled on top (see above) or closed off with a temporary brick-built dome.
In loading the kiln, pottery is sometimes deliberately separated during firing by using ceramic objects called saggers (or setters). Saggers also allowed pottery to be stacked; the saggers containing the pottery sometimes fitted into each other so that a larger number could be stacked. The pots also sometimes stood on kiln plates within the saggers. For obvious reasons the saggers would have been made from a high-firing ceramic materials (see for example the scientific investigations of saggers used in the production of Chinese celadon wares in Section 4.6.5.4 ). It was especially important that glazed pots were separated during firing, and there are many examples (up until very recently) where one can observe the telltale use of tripods (stilts) which leave three small indentations on the underside of the pot; these were used to separate glazed pots (Figure 4.13).
A wide range of updraft kiln structures may well have been used (see V.E. Swan’s comprehensive publication, Pottery Kilns of Roman Britain (Swan 1984: 34–8, Figures II and III)). Swan describes four different kinds of kilns that were used in the Britain during the Roman period (in addition to ‘open’ firing): (i) a single-chambered sunken kiln with a permanent domed superstructure; (ii) a kiln built on the ground surface with open-topped temporary superstructure; (iii) a sunken kiln with a permanent open-topped superstructure; (iv) a sunken kiln with an open-topped superstructure made from a combination of temporary and permanent materials. In addition to this Swan distinguishes between nine different systems of support for the kiln floors which were either perforated floors or an arrangement of radiating bars from a central support (see Figure 4.11, 5–7). Forms of support for the floors included a free-standing pedestal, cross-walls, corbels, pilasters and an inverted pottery vessel. Another shape of updraft kiln is square in plan.
Downdraft kilns are often elongated in plan and the fire is located at one end with the chimney at the other (see Figure 4.14). The fire may or may not be set in a separate chamber. Downdraft kilns differ in one obvious way from updraft kilns: instead of the draught being drawn immediately through the pottery stacked in the upper chamber the draught initially passes around (not through) the pottery, up the inside of the kiln wall and then downwards off the roof of the kiln onto the pots. It then passes through the kiln chamber holding the pottery and is vented through an exterior chimney. It is the chimney which draws the draught through the kiln. The effect of a downward draught is that the development of hot spots in the kiln is minimised. Perhaps the greatest development of downdraft kilns was in the Far East where downdraft kilns were built ‘in series’ as multi-chambered climbing kilns (‘dragon’ kilns), such as at Longquan, China (see Figure 4.15). Here the kiln is built on a slope with the firing chamber at the bottom and a series of kilns built next to each other, linked by openings at the bases of the walls on the uphill side. The draught heat therefore emanates from the fire box and is drawn towards a chimney at the top end passing over the pots, being deflected from the kiln rooves, and then through into the next kiln via the hole in the base of the uphill kiln. Clearly the wind direction was important, whether feeding an updraft or a down-draft kiln, so the kilns would have been constructed with the prevailing wind direction in mind. Pots could be inserted through external openings in the kilns and these openings could also be used to examine the pots being fired. Very large kilns of this kind can take as long as two weeks to fire (Leach 1976: 186).
Figure 4.15 A Chinese dragon kiln constructed on the side of a hill, Longquan, China
(photograph courtesy of Nigel Wood).
The construction of pottery kilns, like glass-melting and metal-smelting furnaces, involves the manufacture of large numbers of bricks which can withstand the temperatures involved; it was therefore just as critical to select clays to make the kiln bricks and kiln saggers with the appropriate properties as it was to select clays to make the pottery. Inevitably the bricks became vitrified by the development of hot spots on the inside of the kiln, with a breakdown of the brick structure. In addition, as with other pyrotechnologies, each time the kilns are heated up the bricks expanded and each time they cooled they contracted. The kilns therefore had a finite lifespan, sometimes needing to be patched up with clay when the walls cracked and sometimes buttressed when the cracking was severe. After all, if a kiln collapsed during a firing the time and energy which had been invested in making the pottery load would be completely lost.
Rye and Evans (1976: 167) found that during a kiln firing the atmosphere of the kiln changed from oxidising just before the addition of new fuel to reducing just after the fuel was added (see Sections 4.3.9 and 4.3.11.3). So this was something taken into account when quite specific pottery glaze colours were required (which might rely on the furnace atmosphere for full colour development), given that the kiln would need to be stoked at regular intervals during the firing. Another very important consideration was the way in which the pottery was stacked inside the kiln (see p. 138). Indeed when dealing with mass-produced wares the arrangement of pottery and saggers inside the kiln was carried out by specialists.
If pottery firings go wrong for some reason the pots may break or may be indented (Figure 4.16), or alternatively it may be impossible to identify their original form. If pottery over-heats badly the clay will begin to bloat and the pot will collapse.
Figure 4.16 Wasters from an Abbasid kiln, al-Raqqa, northern Syria; note the cracks and indentations.
The kind of fuel used is clearly important – whether it is slow- or fast-burning — and the density of one kind of fuel (hard or soft) wood, determines the amount of heat produced. Soft woods burn faster and therefore release their heat at a faster rate – an important consideration in bringing about a quick rise in temperature at the start of a firing (Rhodes 1969: 61). The fuel that can be used includes wood (branches, bark, brush, sawdust), charcoal, peat, palm fronds, reeds, bamboo, coconut husks, agricultural by-products such as straw, chaff, corncobs, cattle or sheep dung and the bones of animals and fish (Cockle 1981: 94, Moorey 1994: 144; Rice 1987: 154). The size of the fuel is also clearly important; logs burn more slowly than twigs so they are often placed below the pottery. Peat, dung, reeds and straw are most appropriate for ‘open’ firings and dung in particular because it burns rapidly and evenly, holding its shape as a glowing ember. Smaller fuel and dung placed on top quickly produces a layer of ash which can insulate the pottery and reduces the heat loss by convection. In Italy in the sixteenth century Cipriano Piccolpasso described the use of straw in kilns in order to produce an initial intense heat (Piccolpasso 1980).
The management of fuel resources is clearly something that is relevant to efficient pottery production, especially if it occurs on a large scale (as discussed in the context of metal production in Section 5.6.2). Writing about the Romano-British pottery industries, Swan (1984: 7) has suggested that areas of woodland, properly managed to grow small and medium-sized trees, would have been essential both for the fuel needs of any larger industrial establishment and for the construction of its associated buildings, such as workshops and stores’. She arrived at the same conclusion, by inference, as Figueral (1992) did when she examined scientifically the evidence of fuel consumption involved in Romano-British iron-smelting: that permanent coppicing must have been involved. In the case of pottery production this must have been true especially for large-scale production which operated over long periods, such as in the Oxfordshire region and at Colchester, for example (Swan 1984: 7). Another perceptive point made by Swan is the possible relationship between kiln design and the type of fuel used. Some of the Romano-British kilns excavated at the Alice Holt/Farnham, New Forest, which were twin-flued kilns had letterbox shaped apertures at the junction of the flue and furnace chambers. These apertures have been interpreted as an indication of the use of fuel with a small diameter (Swan, 1984: 7), although, if this was the case, clearly only a small quantity could have been added relatively slowly at some stage of the firing cycle. There is, however, no question that kiln design and fuel availability must have been linked to both scale of production and demand.
Arguably, fuel was the most ‘expensive’ raw material. Wood selection and charcoal burning are both labour intensive. Depending on the scale of production, it was therefore critical to construct a kiln in which the heat was used effectively and efficiently. Although the pottery was enclosed in a kiln, the initial heating up of the kiln itself (as for any furnace structure) required a lot of heat energy – and this is one reason why it can be more efficient to use a large kiln rather than a small one, in that each time a kiln is loaded with pottery to be fired, the kiln must be cooled down – so the larger the kiln, the larger the volume of pottery loaded into it. Cardew (1969: 182) calculated that even in a large kiln 30% to 40% of heat is lost through radiation, convection and conduction. For open firings the figures get far worse – only 10% of the energy potential of the wood is used (National Academy of Sciences 1980: 28). Indeed the availability of fuel may have had a direct effect on pottery technology: Matson (1971: 74) has even suggested that a shortage of fuel in the early first millennium AD may have been responsible for the lowering of pottery firing temperatures at Seleucia.
The late Iron Age in western and central Europe was a time of great social and economic change. Economic centralisation and an increase in the population are reflected in changes in the strategies of production, trade, exchange and, to some extent, settlement, albeit with a level of continuity from the earlier Iron Age. Although it cannot be claimed that the features which characterise the system of large defended areas called oppida in parts of Europe are also found in Britain – with centralised populations, clearly identifiable production areas in some and occasionally a ‘hierarchy’ of structures – there are, nevertheless, signs that Britain increasingly came within the ambit of what was happening on the continent of Europe. Very large sites, often on valley sides, although of a wide variety of types (Bradley 1984; Cunliffe 1991: 366; Millett 1990: 23), surrounded by banks and ditches and which mainly date from after 50 BC, did exist in Britain. Within some of these, coin mints have been found (Rodwell 1976; Crummy 1980). High status burials tend not to be associated with these oppida and this gives us a clue as to the status of (some of) the sites (Millett 1990: 23). With one or possibly two exceptions, there is not the same evidence for the intensification and centralisation of a range of industries, as found in some European oppida (Collis 1984). The first exception is the site of Hengistbury Head in Dorset. Here excavations have revealed evidence for the working of iron, copper alloy, silver and shale, for salt extraction and for possible glass-working (Cunliffe 1991; 1987). By apparently being at the edge of the oppida system Hengistbury benefited from exchange and trade in a range of materials and products. Those which formed part of the processes of production included raw glass, shale, lead (which was cupelled to produce silver) and copper (Cunliffe 1987). Hengistbury can, however, be regarded as anomalous in a British context; no other late Iron Age site, including those labelled as oppida, have revealed such a range of industries. Indeed in Britain only the earlier Iron Age site of Meare Lake Village, Somerset, can be claimed as producing evidence for such a wide range of industries (Bulleid and Gray 1948; 1953; Gray 1966 and Coles 1987). The other possible exception is the site of Silchester, where excavations of an oppidum have produced evidence of coin moulds (Fulford 1987: 275) and revealed the evidence for the manufacture of copper alloy horse-riding furniture and for iron-working in contexts dating to the first half of the first century BC (Fulford, personal communication). Silchester/Calleva, enclosing 35 ha, may also turn out to provide evidence for industrial production on a scale comparable to that found in continental oppida (Fulford 1987). The oppidum of Stanwick and the associated site of the Tofts in Yorkshire have produced evidence for copper alloy working, including the production of horse-riding furniture, and other copper alloy working, but no evidence for industrial installations (Spratling 1981; Haselgrove, personal communication).
This picture of relatively dispersed (non-centralised) industrial production in late Iron Age Britain and the lack of correlation between large sites and centralised focal places in society (Haselgrove 1986) is further underlined by the evidence for the largest scale of coin production in Iron Age Europe. This has been found in Lincolnshire, at Old Sleaford (Elsdon 1997), where 3,000 coin moulds have been found, and at Dragonby (May 1996). Neither of these sites can, strictly speaking, be classified as oppida. In Europe it is possible to identify incipient urbanism reflected in the features of production industries in what can be described as industrial villages (Collis 1979; 1982; 1984; Pieta 1982). Some heavily-defended European oppida have produced evidence for specialised industries, sometimes on a large scale, such as at the 3 sq km site of Gradistea Muncelului, Rumania (Daicoviciu and Daicoviciu 1963) and at Manching, Bavaria (Piggott 1965: 216; Krämer and Schubert 1970; Jacobi 1974; Collis 1984; Ralston 1988). Although this pattern of industrial centralisation, with the inferred control over raw material supply that this involved, may have been true for areas like Bavaria, it was by no means the case for all of late Iron Age Europe. In central Bohemia, for example, excavations of the late Iron Age settlement of Mšec revealed nineteen iron-smelting furnaces (Pleiner 1977; 1980). These furnaces must have been producing far more iron than was needed for local consumption. Moving westwards, to Burgundy, the range of settlement sizes, including small oppida, that have been observed (Ralston 1988), though not proven to be contemporary, also suggests a lack of centralisation.
In Britain it has been claimed that industrial centralisation occurred in another large settlement type somewhat earlier: fortified hillforts. However, if the evidence for middle and late Iron Age industry in Britain is examined closely, it becomes apparent that a large part of it is to be found outside hillforts in much smaller settlements (Henderson 1984; Henderson 1991: 112; Morris 1994a: 384) which often specialised in single materials (Henderson 1991). A definite exception to this is the distribution of middle Iron Age pottery from hillforts in western England (Morris 1994a: 378). Where it can be shown that hillforts are contemporary with smaller industrial settlements in the same region, the supply of raw materials may have been controlled from these larger sites. However, at the moment this cannot be demonstrated. Hillforts were important settlements in other ways and probably acted as meeting places and grain stores, as has been suggested for the most extensively excavated hillfort at Danebury in Hampshire (Cunliffe 1995). It is, nevertheless, irrefutable at present that evidence for iron-working from hillforts, for example, is scant when compared to the massive amount of evidence which has been discovered at smaller sites which specialised in iron production such as at Gussage-all-Saints, Hampshire, Weelsby Avenue, Lincolnshire, and Bryn-y-Castell, Snowdonia. Compared to pottery, iron is less easy to characterise and it is less easy to build up distribution patterns, but the large scale of production evidence speaks for itself (see, for example, Section 5.12.2.1 ). It is unlikely that iron produced at relatively small settlements was solely for local use. This is in accordance with the production of exotic Iron Age materials, like glass, where the products, even in the middle Iron Age, spread well beyond production zones (Henderson 1991: 125, Figure 4) so it is likely that an excess of iron was produced for local consumption. On the other hand in the late Iron Age, especially after c. 50 BC, a range of pottery fabrics and forms was being produced in Britain and imported from abroad.
Given that the remains of shaft furnaces used for iron-smelting have been found in Iron Age Europe, the technology of their construction and of achieving sufficiently high temperatures, with the necessary fuel/draught supply, was clearly understood. We might therefore expect to find extensive evidence for Iron Age pottery kilns in Britain and in Europe, but up to now no such evidence has been found. This might be because the high temperatures involved in iron-smelting of c. 1300 °C would have led to more durable vitrified ceramics at the end of the smelt than would have resulted from firing pottery. But just because the temperatures necessary for iron-smelting were achieved, this does not by itself indicate a priori that similar developments would have occurred in pottery manufacture. A range of social/political and ritual factors may determine tight norms in the development of industries in which technological aspects of particular specialisations may not have been shared.
One of the fascinating aspects of Iron Age society towards the end of the first millennium BC was the links which existed with the Classical Roman world. Although the ‘Celts’ (referred to as the Gauls by Livy) had invaded Italy and sacked Rome in 387 BC, they also imported very fine ceramic and metal products from the Classical world. While these have been found on high status sites (especially burials) in Europe dating to the fifth and fourth centuries BC, the adoption and use of a higher proportion of Roman products on a regular basis did not really occur until the second century BC. It is no coincidence that this occurred at the same time as the emergence of settlements displaying a range of (different) proto-urban features in Britain and Europe and that increasingly firm trading contacts with the Roman world developed. In no way did Roman technology dominate. Iron Age industries in Europe manufactured some outstanding products, reflecting great skill and understanding of the properties of the raw materials – witness the highly complex modelling in metal of early Celtic art involving the use of the lost-wax technique (Stead 1985; Hodson 1995) the evidence for which has been found at sites like Gussage-all-Saints, Dorset (Foster 1980) and Weelsby Avenue (Foster 1995) (see Figure 4.17).
Figure 4.17 The locations of the principal Iron Age sites mentioned in the text
However, there are clear signs that late Iron Age pottery technology experienced some significant changes from about the first century BC; though on a generally smaller scale, there is evidence that a level of non-local production can be traced back as far as the early and middle Iron Age (fifth–second centuries BC). Peacock (1968; 1969) showed relatively early on in the technological investigation of Iron Age pottery that it was possible to characterise the products of pottery made in south-western Britain because it contained exotic mineral suites which characterised the rocks found there (Peacock 1969: 52–3). The assumption has been that the petrological characterisation of pottery in eastern Britain, where there are not the same deposits of igneous rocks which could provide a characterisation, would prove to be a lot more difficult. The reason for this is that the clays and tempers which are available in eastern Britain are unlikely to be as easily characterised as those in the west because, typically, the shell-tempered wares, which are particularly common in the region, are difficult to characterise due to the occurrence of fossil shells at numerous sites along the Jurassic escarpment (Elsdon 1997: 124). In very few instances, where the nearest source of fossil shell is at some distance from a site a non-local source can be suggested. At Cowbit, Lincolnshire, for example, where the distance is 14 km, regional distribution of the pottery or the shell temper can be suggested in connection with seasonal exploitation of salt at the site (Knight 1999). Petrological analysis of a range of pottery from Lincolnshire sites was carried out (Middleton 1996) in order to examine the relationship between fabric and pottery decorated with specific techniques, which has led to the suggestion that they were specialised products (May 1996: 436; Elsdon 1997: 108). In the event, no clear link was detected. However, the use of neutron activation analysis to investigate pottery from Ancaster and Dragonby did show a distinction in the clays used; it is clear that further investigations using a technique which is capable of detecting elements at low levels may provide a means of characterising the pottery, but not necessarily help to identify analytically a clay source.
In Lincolnshire, therefore, a tacit assumption that it is impossible to characterise late Iron Age pottery using petrological and chemical analysis has been found to be only partially correct: there are promising avenues of research which confound this assumption. Work in the counties of Nottinghamshire, Derbyshire, Leicestershire and Northamptonshire has shown that initially during the middle Iron Age (fifth/fourth–second century BC) pottery was produced which was tempered with igneous inclusions which possibly derived from diorite sources in Charnwood Forest (Leicestershire), or at its fringes – as suggested by Dr John Carney of the British Geological Survey (Allen et al. 1999: 129). This is the same area that is thought to be the source for group XX stone axes (see Chapter 6), some early medieval pottery (Williams and Vince 1997; and see Section 4.5.2.2 ) and its study may provide evidence for ‘trade’ in stone querns (Knight 1992). The coarse angular inclusions are derived from medium-grained rock of granitic texture, further characterised by pink to grey feldspar, abundant grey quartz and flakes of biotite. It was first recognised at Gamston, Nottinghamshire, in quartz-gritted and grog-tempered sherds (Williams 1992; Knight 1992). This granodiorite temper was found to be similar to the outcrop close to Mountsorrel on the eastern part of Charnwood Forest. Knight (1992: 42) has suggested that the pottery or temper was transported over a distance of 35 km from Charnwood to Gamston. Other pottery tempered with granodiorite derived from Charnwood has been found in Derbyshire, Leicestershire and Northamptonshire (Knight 1999). In addition, pottery made from gabbroic clays, typical of the Lizard peninsula in Cornwall, has been found as far afield as Weekley, Northamptonshire, about 380 km distant from the source of gabbroic clays (Knight 1999) and at Crick, Daventry (A. Woodward, personal communication). The same clear distribution has not been achieved for pottery moving in the opposite direction.
There is one clear sign of Roman influence on ceramic technology in the Iron Age and that is the introduction of the potters wheel (Thompson 1982; Morris 1994a: 383; Rigby and Freestone 1997: 57). Whereas the native pottery of Iron Age Britain had been hand-made and had a variable composition including temper of vegetal matter and grog, the results of using a potters wheel are clearly visible among the more sinuous forms, such as pedestalled urns, with the characteristic series of spiral rills on their interiors (see Section 4.3.3). In parts of southern Britain during the first century BC these wheel-made pots formed a high proportion of the assemblages found at Skeleton Green (Partridge 1981), Elm Park House, Ardleigh (Thompson and Barfield 1986), Ursula Taylor Lower School, Bedford (Dawson 1988) and Baldock (Stead and Rigby 1989). The pottery had more sinuous forms compared to the local hand-made wares, clearly as a result of being produced using a wheel (Figure 4.18). Although made using a new technique, the raw materials (grog and vegetal temper) are often indistinguishable from those used to make hand-built traditional wares (ibid.). Production was still very much of a regional character, and occasionally it is possible to characterise the raw materials used to the extent that it is clear that not only ‘local’ raw materials (i.e. those which derived from an area within the vicinity of the site) were used, but a proportion derived from some distance from the site.
Figure 4.18 A wheel-made later Iron Age ovoid jar with a bead rim from Gamston, Nottinghamshire
(after Knight 1992).
An example of this is Hamilton’s (1985) study of seven Iron Age fabrics identified from the excavations at the second–first-century BC late Iron Age farmstead of Copse Farm, Oving, West Sussex, England (Bedwin and Holgate 1985). The pottery encompassed both ‘saucepan’ pottery with a date span of c. late second–early first century BC (Hamilton 1985: 225) and the succeeding Aylesford-Swarling types, traditionally dated to between 50 BC and 43 AD (Cunliffe 1991: 132ff.) with the Copse Farm material likely to date to the late first century BC. The ‘saucepan’ pots were largely barrel-shaped and some had tooled horizontal lines just below the rim and above the base angle; some had more elaborate decoration. The forms of the wheel-thrown Aylesford-Swarling pots consisted of jars and bowls, often with out-turned rims and many had cordoned decoration. An example of a corrugated urn made in the Aylesford-Swarling tradition of Birchall’s type II (Birchall 1965: Figure 4, no. 34) is shown in Figure 4.19. The site produced what have been identified as hand-made copies of Aylesford-Swarling types. In this context it should therefore have been possible to investigate the extent to which fabric-use related to the technique of forming – and whether raw materials used for ‘the more specialised’ wheel-thrown pots derived from further afield than the hand-made pottery. This presupposes (potentially falsely) that a high proportion of wheel-thrown pottery was made locally for ‘local’ consumption. Hamilton found that two fabrics, a wheel-thrown quartz sand-tempered fabric (fabric 3) and a hand-made flint-tempered fabric (fabric 1) represented 55% and 28% respectively of the sherds examined. The quartz sand used for wheel-thrown pottery (fabric 3) was regarded by Hamilton as so fine as to be a ‘glass sand’ and to have derived from up to 7 km away from the site. On the other hand the raw materials used to make hand-made ‘saucepan’ pots may have included clays derived from local Woolwich or Reading beds. However, it is suggested that the flint temper used in these pots may have been coastal pebbles – and these would have derived from about 10 km away. There is no question that fine sand rather than razor-sharp flint temper is more suited to wheel-throwing – the presence of flint would cut into the potter’s hands – and in that sense the change in raw materials does, in this case, reflect a change in technology. Even though there is an absence of evidence for local pottery production, it is still possible to infer that the hand-made pot was made locally due to the petrological identification of the clays used to make pots and mineralogical links to locally-occurring clays. On the other hand the more standardised fabrics used to make the wheel-thrown pottery are more likely to have been imported to Copse Farm. This concurs with the model for a more centralised mode of production, whereby a smaller number of centres produced wheel-thrown pottery, sometimes in finer fabrics, in the late first century BC and early first century AD, anticipating the arrival of Roman production practices.
Figure 4.19 An imported corrugated urn of first century BC date made in the Aylesford-Swarling tradition from Old Sleaford, Lincolnshire of Burchall’s (1965: 245) type II
(after Elsdon 1998).
This could be described as contributing to a greater degree of production specialisation, especially if the raw materials were seen to ‘improve’ the appearance (as may be suggested for wheel-thrown pottery) or of its firing behaviour. Both the wheel-thrown and hand-made pottery were also clearly fired in open firings rather than in kilns. The temperatures did not reach those which were necessary to burn out the vegetal inclusions, suggesting that the firings were short and relatively reducing; no kilns have been found during excavations of British late Iron Age sites. Had kilns been used one might suggest that the workshop mode of production was involved, rather than that of ‘the household’ (Peacock 1982: 8–9).
From about 25 BC a range of wares which reflects the adoption of Romanised dining customs, are found on some British Iron Age sites. In contrast to the locally-made wheel-thrown pottery, these imports include some made with fabrics which can be characterised by the presence of exotic volcanic and metamorphic rocks, suggesting a source for the pottery in central Gaul (Rigby and Freestone 1985). These imports are found in high status burials such as at Welwyn Garden City, Hertfordshire (Stead 1967) and a mirror burial at Dorton, Buckinghamshire (Farley 1983). These wares contrasted with the locally-made British wares in being of an even pink colour, sometimes having mica-dusted coatings and occasionally white slips. They were clearly mass-produced in permanent kilns using a finer fabric and were a far more standardised product. Although these finer wares were available, it is interesting that locally-made wares were still produced throughout the Roman period in Britain.
Rigby and Freestone (1985) note that the pottery forms made in the latest (pre-Roman) Iron Age are typologically and technically identical to those found in the Belgic assemblages of northern France. It is only by using scientific analysis that they can be distinguished. This very close similarity seems to suggest something that is rare to be able to pinpoint in an archaeological context: that the British products were first made by immigrants or itinerant Gallic potters who used locally-occurring raw materials.
In total contrast to the manufacture of pottery made on the wheel, rather coarse hand-made ceramic containers were produced specifically for drying and transporting salt. One of the better-known sources for salt, which continued to be exploited from prehistory into the medieval and later periods, is in Cheshire at sites like Droitwich (Hurst 1997) and Nantwich. Another recent publication of the site excavations in Droitwich (Woodiwiss 1992) reports the discovery of coarse ceramic containers, simple hearths with bars and pedestals, water channels and plank-lined pits used for storage. The evidence published here is both Iron Age and Roman in date and relates to the extraction of salt from saline springs in Droitwich. Some of the evidence that has been found for salt production, at sites like Brean Down, can now be dated to the late Bronze Age, and the known number of Iron Age sites that were involved in salt production is growing all the time. During a survey of the south-western fen edge in an area of approximately 20 sq km, 192 saltings were found, mainly in south-western Lincolnshire (Lane 1992: 218). The dates of these saltings ranged from late Bronze Age (with a carbon-14 date of 810–415 cal. BC at the 95% confidence limit) at Billingborough, to (middle) Iron Age, Roman and medieval. The sites are normally identified by the presence of fired clay fragments and dated by associated pottery, where present. Salt extraction from saline water was normally carried out at low energy inland parts of tidal creeks (saltwater marsh) close to the freshwater fen which could provide peat or wood for fuel. During this extensive survey various roughly-made baked clay artefacts associated with salt extraction were found. Bars, supports and vessels (mainly ‘non-circular’) constituted the principal types (Lane 1992: 221–6). It is suggested that the crudity of the clay indicates that the equipment was produced on site. Some of the clays used to make this briquetage had undergone preparation with the addition of chopped vegetation, including cereal waste from threshing in some cases; such additions have been found also in Hampshire briquetage (Bradley 1975: 23). Further south, on the coast of Essex, by 1990, 315 individual ‘red hill’ (saltings) sites had been discovered, mainly of late Iron Age and Roman date (Fawn et al. 1990). Excavations in the Red Hills produced evidence of settling tanks (to allow the water to clear), and hearths (some of which had become vitrified).
The actual reconstruction of the processes of evaporation and crystallisation of saline water, while simple processes physically, is relatively difficult using the available archaeological evidence of hearths and briquetage. Clearly the saltwater was heated in pans in order to drive off the water. However, it is likely that evaporation would have occurred as a continuous process, with the tank being topped up with saline water before the salt crystals were completely dry. The reason for this is the presence of what medieval producers called ‘bittern’, the dissolved matter which produced a bitter taste in the salt: if all the water was driven from the saline solution the bittern would be mixed in with the salt crystals. It is likely therefore that the salt crystals were scooped out as they formed. The crystals would have been transferred for drying and moulding into cakes. However at the Essex Red Hills sites no suitable candidates for drying hearths have been found, though fragments of suitably low-fired vessels have been.
Studies of salt production in Wessex (Bradley 1975; 1992; Morris 1994b) have provided a model for changes in the intensity and scale of the industry over time. The production of salt on the south coast of Wessex in the earliest known phase during the early–middle Iron Age was seasonal and on a small scale. During the late Iron Age several salt-producing sites appeared on previously uninhabited shores, a reflection of a more intense, larger scale of production for a wider network of consumers. Production was probably part-time, even in the late Iron Age, and it only really ‘came of age’ in the Roman period. The ceramic moulds and other equipment used in the Iron Age salt-producing industry were coarse and presumably produced when the need arose. In the Roman period however permanent inland brine springs, which were already concentrated salt solutions, were exploited and metal cauldrons were used for heating the salt solution (Bradley 1992).
Salt production was a relatively widespread activity but we are still not quite certain about why salt was being produced; nonetheless, one suggestion, that salt was used for drying meat which could then be stored (Cunliffe 1991: 466) must be correct. Because the fabric of the containers which salt was transported in is coarse, it becomes possible to characterise the fabric using thin-section petrology. Morris (1985) in the examination of briquetage from Droitwich and the central Cheshire plain, has found that the inclusions it contains are often sufficiently characteristic to be able to relate it to a production site, or area of production, and that as a result it is possible to build up distribution zones for the material. The briquetage produced in Droitwich, in the vicinity of the saline springs there, was made in two fabrics: one characterised by sandy marl and the other by sandy marl with organic inclusions (ibid.: 346). The briquetage (known as very coarse pottery, VCP) produced on the Cheshire plain was characteristically also of two (different) fabrics. One has a sandy clay matrix containing angular crushed hard rock fragments (the most common being igneous devitrified porphyritic rhyolite and rhyolitic tuff). The other most common type was a fabric characterised by the presence of a microgranite/granophyre rich in alkaline feldspar and poor in plagioclase. Although Nantwich and Middlewich, Cheshire are possible production centres for this ware because the volcanics and granites occur there, there is, at present, no absolute proof for its production at these places.
Bradley (1975) has shown that salt containers made on the coast of the Hampshire-Sussex borderland were distributed up to 60 km from the source, and there is evidence for the distribution of briquetage from inland sites over greater distances. The salt containers which originated at Droitwich, Hereford and Worcestershire, for example, have been found up to 85 km away from a source (Figure 4.20). The pattern produced is referred to as a ‘restricted’ spatial distribution, though it is far from being highly localised. Briquetage made in the Cheshire plain, in the Nantwich area, has been found up to 140 km away (see Figure 4.21). There is also now evidence for the distribution of briquetage from Cheshire further away still, as far as Nottingham in the East Midlands (Knight 1999). Although these distribution patterns provide powerful evidence for a sphere of interaction, the period over which these interactions occurred is very important to the archaeologist because society changed during the middle/late Iron Age, and interactions as reflected by briquetage distribution during different phases of the Iron Age would clearly be articulating with potentially different social structures. Thus, the distribution of VCP during the early Iron Age extends up to 47 km from mid-Cheshire, whereas during the late Iron Age it is distributed as far as Nottingham (Knight 1999a). This is clearly a reflection of the ways in which exchange and trade in such materials operated in a more open system over longer distances in the late Iron Age than in the earlier Iron Age. Morris’s work on salt containers from this source in mid-Cheshire (see Figure 4.22 for material of later Iron Age date) shows that the distribution pattern for the later Iron Age is similar to that found in a down-the-line exchange system (Renfrew 1975). By plotting raw numbers of characterised briquetage fragments against distance from source, one is extracting that particular ceramic from a wider technological context. In order to, in effect, normalise the occurrence of salt briquetage to the occurrence of pottery Morris (1994a: 385–6) uses what she refers to as the ‘Salt index’. This is a calculation of the ratio of salt container fragments to pottery fragments by weight from a site by phase; this accounts for a variety of archaeological factors such as the amount of deposition activity on the site (see Figures 4.23, a–c).
Figure 4.20 Distribution map of later Iron Age Droitwich salt containers and the location of the source
(after Morris 1985).
Figure 4.21 The distribution of Droitwich salt containers and stony VCP (very coarse pottery) showing, by means of pie diagrams the relative proportions and the location of the source of the salt containers at Droitwich
(after Morris 1985).
Figure 4.22 The distribution of later Iron Age Cheshire VCP (very coarse pottery) containers with the location of possible sources
(after Morris 1985).
Figure 4.23a Regression diagram for the distribution of middle–late Iron Age pottery characterised by the presence of Malvernian rock from their source in the Malvern hills, Hereford and Worcester as a percentage of the pottery assemblage on each site (represented by a lozenge) on which they occur in the Severn valley (r = –0.735)
(after Morris 1994).
Figure 4.23b Regression diagram for the distribution of middle–late Iron Age pottery characterised by the presence of Paleozoic limestone from their source near the Woolhope Hills in Hereford and Worcester as a percentage of the pottery assemblage on each site (represented by a lozenge) (r = –0.723)
(after Morris 1994).
Figure 4.23c Regression diagram for the distribution of middle–late Iron Age pottery characterised by the presence of dolerite from their source at the Clee Hills, Shropshire as a percentage of the pottery assemblage on each site (represented by a lozenge)
(after Morris 1994).
Whereas clearly the study of the distribution of Iron Age salt briquetage and its occurrence in relation to the deposition of pottery on archaeological sites is instructive on one level, a further consideration is the extent to which briquetage distributions coincide with other dated Iron Age materials and what this tells us about social groupings. Different materials, such as pottery, coins and glass, will have had different social meaning and socio-economic values. In western Britain Morris’s (1981: 70) petrological study of very coarse pottery shows a distribution which coincides with chemically characterised glass made at Meare Lake Village in Somerset (Henderson 1991: 125, Figure 4), falling as it does within a ‘tribal’ zone attributed to the Dobunni, based on the distributions of Dobunnian coins (Sellwood 1984: Figures 13.1, 13.2). The fact that these three distribution patterns for three very different materials, very coarse pottery, glass and coins, coincide ought to tell us a range of things.
The first point to make is that the range of factors which determines distribution patterns may be wide indeed. Plog (1977: 129), for example, suggests that the range of movement of commodities, the amount exchanged, the time span involved, the direction and intensity of flow, the degree of centralisation of the distribution and the overall complexity of the system are important variables. Although these variables certainly will have played important roles in determining the distribution patterns which we plot for materials, several of these factors, such as ‘the amount exchanged’ and the ‘intensity of the flow’ may sometimes be impossible or very difficult to determine archaeologically (given that we can never be sure that we are dealing with a representative proportion of the amount in circulation or associated with the event which led to the formation of the archaeological deposit yielding the pottery). It is nevertheless well worth examining the data available to see whether these factors can be determined, even if only partially. One of these factors – the intensity of flow – may lead, if it varies over time, to distribution patterns with fluctuating fringes through time according to the intensity of contact, exchange and distribution; the precision of dating techniques available to archaeologists may prevent the detection of these fluctuations, so that what we normally observe is the summed (partial) results of human interaction.
The second point to make is that Iron Age coin distributions are of late Iron Age date because they were first produced in the late Iron Age; the distributions of pottery and glass are of middle Iron Age dates. This provides us with evidence that in this region a putative Dobunnian tribal territory existed some 200 to 150 years before the coins were minted (Henderson 1991: 125). Thirdly, if we regard very coarse Iron Age pottery as having a relatively low value (although its social value may have been potentially high if the salt they contained was used for meat storage), it is perhaps surprising to find an exotic material such as glass falls within the same zone, the materials being distributed from different sources. This would appear to underline how tightly bounded the territory was and, on this occasion, the distribution of very coarse pottery hints at the likelihood that intensity of distribution/contact was quite high. Finally it is worth pointing out that whereas the study of the distribution of petrologically characterised Iron Age pottery contributes in an important way, it is only by comparing its distribution with that of other materials that a more integrated impression of production and distribution can be gained.
In late Iron Age Britain there are a number of factors which contributed to the change in pottery production and which affected the level of specialisation: the availability of raw materials; the use of open firing rather than a permanent kiln; the effect on parts of Britain of the important social and economic changes on the continent of Europe both before and after various areas had been conquered by the Romans in the first century BC; the adoption of Roman drinking vessels – and presumably their dining customs; and local conservatism among both potters and those who used the pots.
The study of other pottery production in the British middle and late Iron Ages has shown that the production of distinctive pottery types became more centralised than during the early Iron Age with pottery being distributed some distance from the production sites. Also there are distinctive regional developments, for example in the west Midlands (Elsdon 1992), with petrological studies successfully revealing complex production and exchange systems in the west country (Peacock 1968; 1969) and the Severn Valley basin (Morris 1981; 1982; 1983; 1991). The result of the comprehensive study of middle–late Iron Age pottery production and distribution by Morris and other workers has provided the evidence for down-the-line exchange systems for traded wares. Figures 4.23a–c show regression analyses for the percentage of traded wares for each site considered characterised by petrology against the distance from the geological source area to sites in different areas. This provides clear evidence for down-the-line exchange, which is apparently independent of the size or form of the site on which it was used. The chemical investigation of the pottery which constitutes the important Iron Age style zones of southern England (Cunliffe 1991) would provide an additional layer of interpretation relating to Arnold’s (1985) resource exploitation territories.
Between c. 410 and 650 the Anglo-Saxons occupied England. They were skilled metalworkers, but they did not continue to make the rather high quality pottery produced by their Roman predecessors. Their hand-made pottery can therefore be described as less technically sophisticated. Most pottery of this date has been found in large cemeteries, with a lesser contribution from settlements. The overwhelming majority of complete vessels come from cremation cemeteries, such as Spong Hill in eastern England, with a smaller number surviving where inhumation burial was practised. When ceramics from burials are examined, rather than those associated with settlements, it is very much a case of studying the ‘way of death’ rather than the way of life and this should be taken into account in any interpretation.
The hand-made early Anglo-Saxon pottery technology involved fabrics which were probably fired at relatively low temperatures, c. 500 °C, or slightly higher, in bonfires (Brisbane 1981: 234). The estimate of this firing temperature was based on the fact that their structurally-bound water had not been lost – for this particular kind of clay (Rice 1987: 103). Most have a black ‘reduced core’ – although this may also be due to the presence of organic materials which were increasingly used in the sixth and seventh centuries (Hamerow 1993: Chapter 3). Pottery was either decorated manually or was stamped (Myres 1969; Briscoe 1981). Aside from the pottery itself, the evidence for the manufacture of early Anglo-Saxon pottery is sparse. Two examples serve to reflect the level of evidence. One piece of evidence is a probable firing area found near Pakenham in Suffolk, where two circular hearths lined with gravel were found. They contained partly baked sherds which had a similar character to clay deposits found within 100 m of the site (Haith 1997: 150). Pot dies, a pottery trial piece and a ‘clay reserve’ were found at West Stow in Suffolk (West 1985). However no remains of kilns have been found there. The implication from this evidence, at least, seems to be that the scale was that of a household industry (Peacock 1982), though there is evidence that this is not necessarily strictly true. There is inferential evidence for more than just a ‘household’ level of industry: the distribution of pots which have a characteristic stamped decoration on them, produced with deer antler, has led to the suggestion that they are the products of the ‘Illington-Lackford workshop’ (Campbell 1982: 36). The distribution of these pots, which occur on ten sites, is in East Anglia and Cambridgeshire. However there is no direct industrial evidence for their manufacture and a study of their fabrics indicates that a variety of clays were used in their manufacture (Welch 1992: 110). This has led to the suggestion that production was indeed on a household scale, perhaps with daughters taking the stamp with them when they moved to a new settlement when they married. Vince’s (1989: 165) view that ‘the low technological level of early Saxon pottery manufacture and its minor role in society has led to a minimal view of its mode of production’ bears investigation in relation to his large-scale petrological research into pottery production in the Thames valley and especially in view of the common use of cremation urns in cremation cemeteries. By the role of pottery in ‘society’ one is referring to pottery used in both functional and ritual contexts.
The geological characteristics of the Thames valley might suggest that a petrological study would not be a promising approach: sedimentary rocks and clays which outcrop in some areas are covered elsewhere mainly by river-lain sand and gravel with mixed rock inclusions from well beyond the Thames valley (Figure 4.24). However, the combination of careful excavation with petrological analysis of large numbers of samples has produced some intriguing results, especially since the study contributes to the interpretation of the growth of proto-urban and urban centres.
Figure 4.24 The distribution of Gault clay and the location of Barrow Hills in the Thames valley, Oxfordshire.
The raw materials for pottery production in the Thames valley included Jurassic Oxford and Kimmeridge clays and Tertiary London clays. These clays contain a very low quartz content and therefore needed to be tempered in order to fire them effectively; this was even more critical when a clay with a high organic component was used which led to an ‘open’ fired texture. ‘Self-tempered’ Cretaceous Gault clays also occur. These contain quartz, muscovite and iron-rich compounds such as hematite and glauconite (Vince 1989: 164). The Gault clays are located at the base and top of the London clay and the Reading beds at the base of the Tertiary deposits. The clay from the Reading beds were used in the high medieval period, but they have a high maturing temperature (the maturing temperature is the maximum hardness and minimum porosity of a particular ware or composition). Thus the potters used either ‘self-tempered’ clays or added temper to clays themselves; the distinction between added temper and natural inclusions can sometimes be difficult to identify using petrology. One kind of temper that can be identified with certainty, however, is chaff, and its presence has been used as evidence for the manufacture of pottery in agricultural communities using locally available raw materials. Research into Anglo-Saxon pottery found in the excavation of settlements has tended to focus on questions relating to domestic pottery production. The small number of decorated pots which have characteristic forms and stamps may be considered to be specialist products, though some plain pots have very regular shapes and some decorated pots have very irregular shapes and are poorly made. In addition fabric differences indicate that some variations in raw materials were involved even for ‘similar’ pots; this could suggest that they were not made in one specialised workshop.
Vince (1989) has carried out an intensive petrological study of early Anglo-Saxon pottery from the site of Barrow Hills, Radley to the north-west of Abingdon, Oxfordshire (see Figure 4.24). More than forty Anglo-Saxon buildings have been found dating to between the fifth and seventh centuries. During the excavations more than 10,000 sherds of pottery were found. Vince studied 3,000 of these sherds petrologically using a binocular microscope and each fabric type was studied more intensively using a petrological microscope. The potential raw materials used, including the gravel terrace on which the settlement was built, and local Kimmeridge clay, were also examined petrologically.
The terrace gravel was found to be typical of that found in the Oxford region, containing abundant rounded fragments of oolitic, fossiliferous and finegrained limestones, a moderate number of fossil shell fragments, sparse red iron compounds and rounded quartz. Lenses of finer forms of silica (sand) are present. It might therefore be anticipated, given that gravel may be heterogeneous, that it would be inappropriate for use as pot temper, though separation of the finer gravel would have been possible and the sands could have been used. There are, however, minor variations in the composition of the terrace gravels depending on its location: those at the foot of the chalk south of Radley contain a noticeable proportion of flint and those in north Wiltshire apparently contain no shell or iron compounds (Vince 1989: 167). These features therefore offered an interesting potential for characterising some of the Barrow Hills pottery. Vince found that the most common fabrics were those tempered with inclusions typical of the Cretaceous Greensands containing quartz grains and some with a glauconitic, micaceous, silty clay matrix typical of Gault clay (see Figure 4.25). These outcrop especially to the east of the site. Other fabrics are typical of gravel deposits in Warwickshire or north Oxfordshire: crushed calcite characterised one fabric, abundant phosphate fragments another.
Figure 4.25 Photomicrograph of a thin-section of early Anglo-Saxon ware from Barrow Hills, Oxfordshire, Characterised by limestone, iron ore and quartz inclusion
(photographed at × 40).
Vince found that the Barrow Hills pottery fabrics did contain gravel terrace temper and that the types of inclusion varied, some with elevated levels of quartz and some with sparse to moderate levels of flint. Although impossible to state where the gravel temper derived from precisely, the important inference is that the gravel temper (and therefore the pottery) was non-local, the assumption being that the temper did not travel and that the pottery was made where the temper occurred. The pottery industry at Barrow Hills was not purely a domestic craft using locally-occurring raw materials; the pottery was clearly made in a supra-local zone, some distance beyond the current parish boundaries of Radley. The question of where precisely this range of pottery fabrics used at Barrow Hills was made is still an open one because the outcrops that were used are widespread, although it would appear that several places were involved.
The vessels are likely to have travelled over both short and long distances to reach the site. The majority travelled over distances of more than 10 km and some for up to 40–50 km to reach Barrow Hills. Trade, market exchange and other mechanisms can be suggested as possible explanations for the distribution patterns of early Anglo-Saxon pottery around Barrow Hills. Carrying this further Vince (1989: 168) has argued that since the settlement was reliant on non-local supplies of pottery it is likely that it relied on non-local sources for other materials; a surplus would have been produced for trade. The research, which has benefited (by a process of elimination) from the use of identifiably non-local temper to make pottery, has therefore provided a model for pottery production in other areas of the early Anglo-Saxon world.
A comprehensive consideration of granitic-tempered pottery by Williams and Vince (1997) using a combination of thin-section petrology and ICPS has also now shown that early to middle Saxon wares with a particular distinctive granitic inclusion are widely distributed. The most likely source of the inclusions in much of the pottery is an area focused on the Mountsorrel granodiorite outcrop in Leicestershire (Allen et al. 1999: 129) with a possible limit at the sand deposits in the Coventry and Warwick areas. This basically coincides with the pre-Cambrian rocks of the Charnwood Forest area; Mountsorrel is located east of Charnwood Forest. (There is also the possibility that the parent rock may have been moved by glacial action.) It is of interest to note that such inclusions were also used in Iron Age pottery from Nottinghamshire and Derbyshire (Williams 1992) showing how long the source of a pottery raw material can be used for (see Section 4.4.2 above). A petrological thin-section of the pottery revealed a clay matrix which is dominated by large discrete grains of plagioclase feldspar, quartz, potash feldspar, biotite, mica and with occasional small pieces of granitic rock composed of quartz, feldspar (predominantly plagioclase but with some orthoclase) and brown biotite’. Occasional grains of other minerals were also present (Williams and Vince 1997: 218).
Williams and Vince found that the ‘Charnwood Forest’ wares were distributed widely in the Midlands with granitic inclusions coming from a ‘single’ source. This therefore provides another example to add to the results of the Barrow Hills study which shows that (some) early Saxon pottery was definitely not produced within a ‘domestic’ local mode of production since it is widely distributed and not simply made for local consumption. Indeed ‘Charnwood ware’ was used both in domestic and religious contexts; Williams and Vince suggest that the pottery may have been exchanged and distributed on the occasion of religious festivals.
Russel (1985: 130) had difficulty classifying Anglo-Saxon plain pottery from West Stow using petrological analysis, finding that identified ‘groupings’ of fabric types tended to be somewhat subjective and that statistical analysis (Clustan) did not help. At Mucking the thin-section programme also experienced difficulties in that in some cases sherds which had been assigned separate petrological types were found to join, again suggesting that the process of assigning petrological groupings was somewhat subjective (Hamerow 1993: 27). Blinkhorn (1997: 118), however, has noted that a broad distinction can be made between chaff-tempered ware and fabrics characterised by containing mineral temper (mica and haematite) as well as ‘chalk/limestone’ and sandstone. By plotting their occurrence at Mucking he found that the greatest amounts of chaff-tempered wares occurred in the northern area of the site. His interpretation of the distributions of decorated wares is that they are partly a reflection of a retention of cultural identity of the Germanic settlers in England; many of the styles of pottery decoration are similar to those found in their continental homelands (Myres 1977). Blinkhorn (1997) notes that although the decoration developed along a divergent path over the next hundred years or so, the pottery remains within the Germanic tradition, presumably indicating a strong retention of cultural identity’.
In his study of 7,000 sherds of early/middle Saxon wares from Raunds, Northamptonshire, Blinkhorn identified three broad fabric classes (bringing together ten fabric types) based on the treatment of sand, grit or organic temper, added. The distribution of these fabric classes revealed that the sand-tempered wares were more common on the part of the North Raunds site where there was a focus of middle Saxon activity; the inference being that although both quartz- and sand-tempered wares were used in the early Saxon period when the site was founded, by the middle Saxon period the less time-consuming use of sand, which did not need crushing, was adopted. This appeared to be a sensible functional interpretation.
Thirty kilometres away the Saxon site of Penny-land, Milton Keynes (Blinkhorn 1993) produced pottery fabrics which were very similar due to the use of similar geological raw materials. However, in spite of the use of both sand- and grit-tempered wares in the early Saxon period, the opposite situation to that found at Raunds was discovered to have occurred with the use of the grit-tempered wares rising from 51% of the plain pottery assemblage in the early occupation phase, to 72% by the latest occupation phase. This evidence therefore clearly contradicts the functional interpretation of pottery-use at Raunds. Blinkhorns interpretation of this pattern of production and use is a combination of a continuing use of continental traditional practices and that the differing tempering techniques were used by people of differing cultural origins who continued to used the accepted procedures established in their homelands. He even feels that the distributions of pottery produced using a variety of temper is a reflection of ‘the rise and fall of the populations of the various cultural groups within the settlement’.
Blinkhorn’s discussion of the petrological analysis of Anglo-Saxon pottery therefore provides us with an intriguing level of interpretation, indicating that a close scrutiny of the distribution of petrologically characterised ‘plain’ wares can generate more than purely a functional interpretation, and indeed, taking residuality in the archaeological record into account, the functional interpretation in this case appears not to make sense. At the very least this approach reminds us that there are cases where functionality must have played an important part in the selection of ceramic raw materials, but that in some cases what was perceived as the cultural norm may have played a more important role in determining the raw materials and other procedures used. It would be very interesting to investigate the use of temper in Anglo-Saxon pottery in the region which includes Pennyland and Milton Keynes, and how any patterns that would be discerned, could link with those established. Clearly there must also have been situations where the cultural norm and functional considerations were one and the same thing.
Most Thames valley settlements including Barrow Hills and Mucking were abandoned in the seventh century. Since mid-Saxon rural settlements in the Thames valley are scarce, Vince turned to an important gravel terrace urban site, London, to investigate pottery production (see Figure 4.26). Unlike the area around Barrow Hills the gravel terraces of the Thames in the London area are capped in places with brick earth, and under the gravel is fine-textured London clay. This sequence would have been visible to potters looking for raw materials in the sides of river and stream valleys. Blue clays were also forming under anaerobic conditions at the time.
Figure 4.26 Location of middle Saxon sites in the London area of the Thames valley.
As in the early Saxon period, chaff-tempered wares were manufactured and commonly used during the mid-Saxon period along with Ipswich-type ware found throughout East Anglia and other wares such as northern and southern Maxey-type wares in Lincolnshire, Northamptonshire and Cambridgeshire. Imported wares came from the Rhineland, northern France and Belgium. Using petrological analysis Vince (1989: 169) found that the chaff-tempered wares contained quartz, muscovite and iron compounds, but these minerals, individually or in combination, are not sufficiently distinctive to be able to characterise the wares to a source. On the other hand the wares containing temper of rounded quartz with polished surfaces are similar to the temper used in Barrow Hills wares; the London quartz temper differs from the Barrow Hills quartz temper because it is covered with a layer of red iron ochre which extends into the cracks of the quartz grains and remains visible even if the grain is polished. Cretaceous Greensand is likely to be one possible source for this quartz temper in the London wares because the Greensand is cemented with iron to form an ironstone. Another is where the Woburn sand outcrops in Buckinghamshire and south Bedfordshire.
The hinges of shells which survive in pottery can provide an interesting means of identifying the species of the shell (Cooper 1982) if thin-sectioned. A small group of London mid-Saxon wares contain bivalve mollusc shell fragments which would be characteristic of the littoral zone (see Figure 4.27). These can be further characterised by the presence of opaque streaks running parallel to the surface of the shell caused by iron pyrite crystals secreted by the mollusc. It might be suggested that this characteristic when found in pottery could be used to link the production of the pottery to a quite specific source of clay containing such shells. However, pottery containing fragments of such shells not only occurs in London, but also in Beverley (east Yorkshire), York and Waltham Abbey (Essex), as well as at Dorestad (Holland) and Flanders. Such a wide distribution suggests otherwise. Research by Hamerow, Hollevoet and Vince (1994) has shown that a typical seventh-century fabric (chaff-tempered ware), with a highest occurrence in England south of the Thames is also found in substantial quantities in Flanders. It is considered most likely that this pottery was copied in Flanders as a parallel technological development (ibid.: 16); on this occasion it has been found that thin-section petrology has not proved to be sufficiently sensitive to establish whether indeed the pottery from Flanders was made there or in southern England.
Figure 4.27 Photomicrograph of a thin-section of middle Saxon Shelly ware from Jubilee Hall, London characterised by abundant shell fragments
(photographed at × 40).
A single pottery sherd which is characterised by the presence of biotite and igneous rock fragments (together with limestone) suggests a source in the east Midlands where the Charnwood Forest granite and Croft syenite outcrop – the gravels surrounding the area – have these characteristic inclusions.
Unlike the study of Barrow Hills pottery the use of petrology in the investigation of the production and distribution of mid-Saxon pottery is therefore rather mixed; at least for Barrow Hills wares it was possible to contribute in a substantive way to new models of pottery production and distribution. One reason is that the London gravels and brickearths are less easy to characterise than those on which Barrow Hills was located. In spite of this, it has been possible to state that some ‘fineware’ pottery was probably obtained from the continent and that coarsewares were obtained from as far afield as the east Midlands – though at this stage the relative numbers of samples are rather limited. Vince notes (1989: 171) that the distances over which pottery travelled are comparable to those found in the late Roman and high medieval periods and that because the appearance of mid-Saxon pottery is somewhat unsophisticated, there has been a preconception that its supply occurred over relatively short distances. The suggestion is (ibid.) that the pottery distribution is symptomatic of an integrated economy in which rural goods were traded for imported goods (of which the continental pottery is an indication). This evidence concurs with the mid-eighth-century coin distribution from the London mint and with surviving documentary evidence. Although clearly in such circumstances there may be strong indications that there is evidence for an integrated economy using coins, if an entirely different level of evidence, such as the production and distribution of pottery agrees with it, the level of interpretation becomes an increasingly more powerful one.
Looking again beyond the shores of England, the Saxon ‘Charnwood Forest’ pottery described by Williams and Vince (1997) has been found to have a very similar fabric and surface treatment to that used in Scandinavia and the Baltic coastlands between the fifth and the ninth centuries. Thin-section petrology was unable to distinguish between the wares found in the two areas, so ICPS was used as an investigative tool instead. Analyses of seventh- to tenth-century pottery samples of this granitic-tempered ware showed that sherds found in Poland could be distinguished from some found in the Malaren valley in Sweden and from Catholme, Derbyshire; results from the Birka, Sweden, material, however, are less easy to interpret. In the main it seems that the Baltic and English granitic wares could be distinguished. However, no information is provided about what elements cause the clustering of ICPS data, and how it might, or might not, be relatable to characteristics observable in thin-section.
The discovery of coin hoards and Scandinavian artefacts in the ninth century are a testament to the fact that London was raided a number of times by the Vikings. The Saxons built timber fortresses at a number of locations along the Thames valley to protect themselves: at Oxford, Wallingford, on an island near Cookham called Sashes, Southwark and London. These places were towns which controlled trade and from the mid-tenth century they had their own mints, which later provided danegeld. By the time the Domesday Book was written, about 10% of the population lived in towns and was engaged in trade or providing services to other members of the urban population. The economy of the mid-eleventh century provided the foundations for the high medieval period.
To what extent is this clear development towards a fully integrated medieval economy reflected in the production and distribution of late Saxon and early medieval pottery in the Thames valley? The pottery which characterised the mid-Saxon period was replaced by another (different) kind of shelly ware: a fabric which contains fossil shell fragments. The ware is widely distributed along the Thames valley and northwards as far as the west Midlands; in spite of having the same appearance and petrology in a range of areas it is known variously as Stafford-type ware, Cheddar E ware, Lincoln ware, Oxford B ware and late Saxon Shelly (LSS) ware. Vince examined forty-five thin-sections of pottery which derived from London and five samples from Oxford. All were characterised by the occurrence of fragments of an oyster-like shell of the species gryphaea which is characteristic of the Oxford clay (Vince 1989: 173). The presence of a single species of shell may be due to its robustness following partial weathering, or because marginal habitats are prone to have a more restricted species range, but as yet the exact clay source(s) has not been located, so neither hypothesis can be proven. The occurrence in LSS of shell fragments of a ‘recent’ freshwater species and sparse rounded quartz pebbles several millimetres across suggests that deposition in a recent alluvium has occurred.
Although widely distributed, there are problems dating LSS. For example, Oxford B came into use in the late eighth century to early ninth century, though neither carbon-14 dating and especially not thermoluminescence are able to provide sufficiently accurate dates to better other kinds of (archaeological-stratigraphic) dating. The pottery probably ceased being used in the eleventh century when Oxford was sacked: dendrochronological dates for the London waterfront, a far more precise dating technique, when available, has shown that LSS formed more than half the assemblage in use c. 1040 but had gone out of use c. 1055. With a wider range of pottery fabrics, and in spite of some problems of generating the necessary decade-by-decade dating framework, it would have been possible to suggest how the production of pottery and the mechanisms for its distribution changed through this important period for the Thames valley. However, at present this is not the case. As the dating framework becomes tighter it will become possible to at least sample LSS from well-dated contexts, which may lead to the recognition of subtle variations in its production technology and use over time. Since LSS is found on all site types – the centres of Domesday vills, towns and minor settlements – the identification of closely-dated technological variations and the changing distribution patterns that can be assembled using these identifications, are potentially very important archaeologically in providing evidence for spheres of interaction using pottery distributions.
In contrast to the late Saxon ‘monopoly’ of wares, during the course of the eleventh century a range of other wares replaced LSS in the Thames valley. At this time the ware types begin to show greater concentrations around towns like Oxford and London rather than the same ware being distributed along the Thames valley; continental imports start to reappear. Vince has also studied petrologically these early medieval coarsewares. He was able to show that in two wares the potters used the same temper as in the early Saxon period: Oxford AC contains limestone gravel which matches precisely the limestone gravels of Barrow Hills and elsewhere around Oxford’ (Vince 1989: 174). The pottery used in London known at the time as early Surrey coarseware was characterised by Vince (ibid.) as containing iron-coated quartz and iron-rich cement fragments of Greensand origin. The balance of mid- to late-eleventh-century coarsewares used in London have variable temper containing mainly well-sorted quartz sand, some with mainly fossil shells from the Woolwich beds, and some a mixture of the two. The same forms and fabrics are also found in north-east Kent; it is no coincidence that the distribution of a fabric with these characteristics coincides with a suitable parent rock.
Another coarseware is characterised by calcareous blue-green algae, originally mistakenly identified as chalk and mistakenly labelled early medieval chalky ware’. The fossiliferous algae were accompanied by polished rounded quartz grains, angular chert and flint sitting in a micaceous, silty clay matrix (see Figure 4.28). The presence of algae suggests strongly that the clay was recently deposited; the presence of freshwater pea mussels supports this. The distribution of these fabrics is mainly to the north-west of London in St Alban’s, Aylesbury and Cublington, so the suggestion is that the source of clay lies in that area – the most likely parent clay being Gault (Vince 1989: 175).
Figure 4.28 Photomicrograph of a thin-section of London early medieval ‘chalky’ ware characterised by inclusions of blue-green algae (× 40).
The use of ceramic petrology alone to examine early medieval pottery in the Thames valley has provided results which contribute to the discussion of how local and regional economies developed during the second half of the first millennium AD. Although examination of the fabrics which derived from the area of Barrow Hills in the Thames valley did not provide a suite of exotic minerals which might unambiguously provide a clay source, by examining very large numbers of sherds it has become possible to show that they were probably non-local in origin. Although a survey of the distribution of gravel deposits revealed where they might have been ‘dropped’ by glacial action, a consideration of possible re-deposition by river action has not apparently been considered. Nevertheless, taking into account that a proportion of the gravels would not have been affected by river action, this is an important result. A further qualification is that although grain size and angularity of inclusions have been taken into account, the relative proportions of different materials used in the pots were not described. This is far from being a severe criticism, since the overall results are perfectly valid, but it would be interesting to examine thin-sections of the pottery with this in mind.
This research demonstrates that the tacit assumption of the ‘household’ production mode for pottery, made within the settlement for consumption there, is now no longer tenable for all of the pottery of the early Saxon period. This conclusion has been supported further by analytical research of early medieval pottery in the west Midlands based on a different means of characterising the pottery, the occurrence of ‘exotic’ inclusions. Indeed, by examining other wares in the Thames valley from mid-Saxon, late Saxon and early medieval periods it is clear that not only the sources of clays changed through time, but that a proportion of the wares were imported from a zone outside 10 km in all the periods considered. It is also interesting to note that the same temper was used in early Saxon coarsewares found at Barrow Hills, Oxfordshire, as in early medieval coarsewares (Oxford AC) found in Oxford, and in this case this observation could be a result of the gravel having been moved by river action. Clearly, as the population increased, and the fully-fledged market economy of the high medieval period came into existence, pottery tended to be exchanged or traded over longer distances; wares made in England may have been exported to Flanders in the seventh century, though this is difficult to prove (Hamerow et al. 1994). Vince’s study has shown several things: first, how important it can be to examine large numbers of samples, when they are available; second, how important it is to frame the research questions so that it has the greatest potential for contributing to mainstream archaeology; and third, the value of using several different (technical and archaeological) approaches in the investigation of the same problem.
The production of Yaozhou celadons will be described here in detail because comprehensive archaeological research has revealed one of the most complete series of ninth- to eleventh-century workshops and kilns ever found in the ancient world. To be able to relate the excavated workshops to the kilns they almost certainly supplied with unfired pots is unique. In addition, few researchers have compared the technologies of the ceramic materials used to make kiln materials, such as saggers, with the pottery itself. The scientific research which has been carried out has gone some way in helping to define the degree of industrial specialisation and refinement. In this instance, both had clearly attained a high level. However, before these fascinating archaeological and scientific investigations are presented, the technological context onto which celadons fit will be described.
Celadons are known in the west by that name because of their green glaze; their bodies are made of stoneware. Before discussing in detail the excavated evidence for celadon production and its scientific investigation, it is useful to consider where celadon ‘sits’ in the developments which occurred in Oriental high-fired ceramics. Amongst oriental ceramic products one can include low-fired pottery, glazed and vitrified pottery (stonepaste wares) produced at higher temperatures, and translucent porcelain produced at still higher temperatures of c. 1280–1400 °C. What is known as Chinese proto-porcelain was first manufactured during the Shang Dynasty – seventeenth–eleventh centuries BC (Chen Tiemei et al. 1999: 1003) using ‘porcelain stone’ (petuntze, china stone), a sericitised or kaolinised feldspathic rock containing kaolinite, muscovite, illite and quartz which can apparently be fired directly into porcelain (Guo Yangi 1987), although work by Yap and Younan Hua (1995) suggests otherwise; one such production centre may have been Wucheng in south-east China (Chen Tiemei et al. 1999). However, it is still not clear if Shang porcelain was ancestral to eastern Han Dynasty porcelain (221–25 BC) (Medley 1986). According to Li Jianzhi (1986: 129–33) the first true white porcelains were being manufactured c. 575 AD at the Gongxian site in Henan province at temperatures of around 1350 °C.
True porcelain is made from high-firing refractory kaolin clays (‘china clays’) mixed with a (partially decomposed) feldspathic rock and quartz. The alkali feldspar melts and fuses to the quartz and also vitrifies the aluminium-rich clay (acting as a flux), so these ceramics, in the same way as Islamic ‘fritware’ and Iznik (see Section 4.7.8) rely for their strength, in part, on having vitrified bodies. The northern Chinese kaolin deposits which were used to make Ding ware porcelain bodies are sedimentary and lack the alkali fluxes which promote vitrification (Guo Yangi 1987) and there is not the same translucency which is associated with the formation of glass in the bodies. The southern Jingdezhen products used kaolin which occurred in association with rocks composed of quartz, mica (sericite) and feldspar (albite – a soda-lime type). The presence of feldspar-rich rocks provided the flux which led to the glassy phase which produced the translucency seen in the famous Jingdezhen porcelains dating from as early as the Five Dynasties period (907–60 AD). The characteristic white colour of true Chinese porcelain can partly be attributed to the purification of the clay used: in the eighteenth century the potters of the famous production centre of Jingdezhen in Jiangzi province used kaolin from sixty miles away which was washed, crushed and re-washed several times. It was then strained through a horse hair sieve and through a bag made out of a double thickness of silk (Staehelin 1965: 22–6).
Stonewares generally fire above 1150 °C and are characterised by having an opaque glassy body. As a result of the vitrification that occurs in the body at these temperatures their bodies have a high density and a low porosity. They are noted for their strength and are generally made from clays which are low in fluxes (like alkalis, iron and calcium) and high in aluminia. Because low levels of flux are present in the clays used, the wares fire at high temperatures, and mullite crystals (an aluminium silicate) form which provide the strength. Celadons are examples of stonewares.
The composition of southern Chinese celadon bodies are characterised by containing relatively high quartz and relatively low aluminia levels when compared to Ding and other northern Chinese white-wares (Pollard and Hatcher 1994). The Yue bodies contain iron oxide which causes them to turn a grey colour when fired and the maturing temperature of southern Chinese celadons is relatively low in comparison with northern wares. There appears to have been a relatively simple recipe for making the bodies using a pulverised altered igneous rock called porcelain stone. This stone consists of clay and fine-grained mica (sericite) – a hydrous silicate of potassium and aluminium (see above). The potassium, although at relatively low levels in Yue ware bodies, for example, acts as a flux and, in addition to the clay, provides the necessary plasticity to form the pot. It is suggested that the lower levels of flux in northern Ding (porcelain) whitewares is due to the use of kaolins. The presence of the flux promotes vitrification of the bodies which gives them a translucent quality.
Scientific and archaeological investigations of the earliest celadons in Korea of the Koryo dynasty (918–1392 AD) have provided evidence that, while the appearance of the pottery itself was clearly inspired by Chinese Yue technology in the late tenth century (Vandiver et al. 1989: 375; Koh Choo et al. 1999: 54), the technology was locally developed (Vandiver et al. 1989: 348, 365; Koh Choo et al. 1999). The kilns were made of mud and rocks: this unusual combination of raw materials was revealed by chemical and mineralogical analysis. There is primary archaeological evidence for the development from the use of bricks as found in Chinese kilns to the use of kilns made from mud and rocks at Söri in Korea. Here a 40 m long brick kiln has been found stratified beneath several mud and rock kilns. Microstructural and chemical analysis of Korean celadons suggest that they were probably fired at relatively low temperatures (1050–1150 °C) and cooled rapidly. There is, however, evidence for a two-step firing method producing a biscuit-fired body for Koreo celadons and this technique clearly distinguishes Koreo from Chinese production methods (Koh Choo et al. 1999: 64). The Korean (Koreo) ceramics were fired in hill-climbing dragon-type kilns (see Figure 4.15) of about 1 m wide and 7–17 m long (Soontaek Choi-Bae 1984) and were smaller than the southern Chinese type. The earlier types consisted of a fire box which was located at the base with stoke holes at one side; exhaust gases and smoke escaped from the top. The hot exhaust and flue air rises to fire the pottery. Later Korean dragon kilns were constructed with separate firing chambers.
The Chinese were involved in making green glazed ware for a span of some 3,500 years from the middle of the Shang Dynasty (c. 1500–1050 BC). Given the generic term green glaze’ there are subtle differences in the colours which need to be explained. Apart from the iron present which clearly contributes in a significant way to the greenware glaze colour (a range of 0.65–1.36% in Guan wares) another potential colorant is manganese oxide, but this generally occurs at low levels, of less that 0.1%. As mentioned in Section 4.3.5.2 an iron-green colour in glass and glazes can usually be attributed to ionic forms of iron, and normally to the presence of the reduced form, ferrous iron (Fe2+) (Portal 1997: 103). However, not only is it likely that, in fact, the iron is normally present as a mixture of two valances, ferrous and ferric (Fe3+), with a higher proportion of ferrous ions in some celadons, but the chemical environment in which it sits needs to be considered (see Section 4.3.5.2 ). For instance the total level of alkali in the glaze has an important effect because the higher the proportion of potassium oxide as opposed to soda, the greater the amount of light absorption occuring and hence the darker the hue.
In fact consideration of the atomic structure in silicate glazes reveals just how complex the coloration of transition metal ions, like those of iron, can be. When a transition metal assumes an ionic state, the ligand (the ions surrounding the colorant) depends on both the field strength and negative charge (i.e. provided by the oxygen). In transition metals, one of the energy shells (in this case the 3d sub-shell) is only partly filled with electrons and it is this which produces some of the colouring characteristics. When co-ordinated with other ions, such as Si4+ and Al3+ in celadons, the energy levels of the d electrons in transition metals are split (distorted) by the electric field produced by the co-ordinating ions. This splitting is sensitive to the arrangement of surrounding ions (the chemical environment); the result determines the glass colour. The theory of these effects is called the ‘ligand field’ theory. When higher energy level orbits in the iron ions are unoccupied, the electrons in lower energy level orbits absorb different wavelengths of light quanta in order to move up to the higher energy level quanta. It is this last energy transition that causes the glaze to appear green: the negative charge is bigger on the Fe3+ ligand than on Fe2+ and Fe3+ and this changes the quantum light resulting in a big absorption in the ultraviolet area.
Naturally other potential colorants make an important contribution. Glazes made at Longquan and Yaozhou have a yellowish brown colour, probably due to the formation of an iron-sulphur chromophore in the glaze (Scheurs and Brill 1984), which would occur under reducing conditions. In celadons made at Yue, translucency is augmented by the precipitation of white anorthite crystals at the glaze–body interface which screens the underlying grey body; sometimes anorthite crystals are precipitated in the glaze itself in areas which are rich in aluminium and potassium oxides (Vandiver et al. 1989: 358; Li Wenchao et al. 1992). The sub-micron sized particles add to the glaze brilliance by bending and scattering the light. In addition, because of intentional poor mixing of coarsely ground raw materials used in Guan and Longquan green glazes, residual raw materials which formed the original batch are frozen in place due to the incomplete fusion. Adding further to the scattering of light from beneath the glaze surface, wollastonite crystals were precipitated in calcium oxide-rich areas of the glazes during cooling in Longqaun celadon glazes of the Song Dynasty (Vandiver and Kingery 1984: 615).
Any crazing which may have resulted from a poor glaze fit in Yue and Guan ceramic glazes may also produce an internal reflection of light which results in a ‘shimmering’ effect. As described in Section 4.3.5, crazing of glaze, in which a series of cracks appear, results from a mismatch in the contraction of body and glaze as they cool together after being fired. Although glaze is characterised by a certain tensile strength, when crazing occurs it is exceeded. The relatively high lime composition of Chinese celadon glazes means that they contract faster than the bodies of the pots. Below 960 °C the glaze behaves like a brittle solid because it ceases to flow or relax. As with any glassy material if, during the period when the glaze is becoming increasingly viscous, the cooling rate is reduced, the amount of crackling may be reduced.
The relatively low-firing Yue glazes tend to be well mixed, finely ground and homogeneous; Guan and Longquan glazes tend to be heterogeneous. The more highly fired northern Yaozhou glazes tend to be homogeneous.
Although Chinese celadon glazes clearly have visual features in common, there are differences in the glaze technologies used in the southern and northern traditions. Among the southern wares (Guan, Yue and Longquan), for instance, Guan glazes have major components of relatively low silica with high aluminia and calcium oxide (with respective mean values of 61.54%, 16.56% and 13.96%). Their alkali (soda and potassium oxide) component is low, with respective mean values of 0.19% and 3.35% (Vandiver et al. 1989: Table III, including samples from Tang, Five Dynasties and Song; see Table 4.1). Of the three southern celadon glazes the Yue wares, in general, contain the highest calcium oxide levels irrespective of date, with the levels of Guan approaching those of Yue; some Longquan celadon glazes on the other hand approach the low levels of calcium oxide found in northern celadons, which tend to contain more magnesia and titania. Yap and Younan Hua (1995) performed a Principle Components analysis on nine greenware glazes – Yue, Longquan, southern Song, Guan, Ru, Linru, Jun, Yaozhou and Ge. In this statistical analysis they included the relative oxide levels of silicon, aluminium, iron, calcium, magnesia, potassium and sodium in the glazes. They discovered that the northern (Ru, Yaozhou, Jun and Linru) clustered separately but close together and that most of the southern glazes could be distinguished from the northern glazes inferring, not entirely unexpectedly, that different raw materials were used in the two regions to make glazes. They were unable to separate some of the southern products using this form of statistical analysis.
Table 4.1 The principal Chinese dynasties referred to in the text
| Southern dynasties of China | |
| Tang | 618–907 (north and south) |
| Five Dynasties | 907–960 |
| Song Dynasty | |
| Northern Song | 960–1126 (north and south) |
| Southern Song | 1127–1279 (south China) |
| Jin | 1115–1234 (northern China) |
| Yuan | 1279–1368 (north and south) |
Chinese celadon glazes tend to be relatively resistant to weathering, although Yue and Guan glazes contain relatively low silica and high levels of calcium oxide (c. 60% silica and 18–24% calcium oxide) and tend to weather to an iron-stained brown colour; any craze lines allowing the water to penetrate.
Re-firing of millimetre-sized pieces of celadon glazes makes it possible to estimate the temperatures at which they were fired, mainly from their behaviour and appearance. Some of these determinations are for kiln wasters, which, in general, tend to be over-fired. Re-firing of southern Chinese Guan and Yue ware glazes between 1000 °C and 1100 °C shows that they rounded at this temperature, with a probable firing temperature of around 1100 °C; this produces the desired translucent effect in the glaze. Above c. 1200 °C the translucency of the glaze is lost because the bubbles in the glaze become larger. These relatively low firing temperatures, a slightly lower total flux and inclusions of unmelted raw materials all limit the degree to which the Guan celadon glazes flow. However, Longquan ware celadon glaze appears to have had slightly higher optimal firing temperatures of between 1100 °C and 1200 °C. Still higher temperatures were used to fire northern Chinese Yaozhou, Ding and Jun celadon wares – optimal temperatures of between 1150 °C and 1250 °C were used; Guo and Li (1986: 157) state that 1300 °C was used. On the other hand, Yue wares tended to be fired at 1100 °C, have pinholes and few large bubbles, indicating that they had a low viscosity.
At the kiln complex of Yaozhou in Hungbao township at Tongchuan (‘Bronze valley’) in Shaanxi Province northern China (see Figure 4.29), some quite exceptional evidence for the production of celadon has been found. During the early and middle Tang Dynasty this complex supplied the nearby capital of China, Changan, at a time when its population exceeded one million. Not only has a series of kilns been unearthed at Yaozhou, but the excavators have discovered the workshops which were associated with the kilns, together with evidence for the processing of clay, emplacements for the wheel and evidence of producing moulded decoration. This high level of evidence is very rare indeed in a world context.
Figure 4.29 Map of Chinese provinces, and principal sites mentioned in the text.
Although the evidence for pottery production at Yaozhou stretches back to Neolithic times, the most comprehensive evidence starts in the late Tang Dynasty (618–907 AD), continued in the Five Dynasties (907–960) and flourished in the years of Xining (1068–1077) in the Northern Song Dynasty and continued in the Jin (1115–1234) and Yuan (1279–1368) dynasties. The celadon made there was hard and eggshell-thin with elegant decoration under the characteristic olive green glaze colour.
Since the evidence for pottery production is unusually comprehensive, it is worth considering here the chronological development of the evidence, which starts with the production of Tang Dynasty glazed wares. It is suggested that these mantou kilns were used for firing lead-glazed porcelain, but there is no apparent evidence that this is definitely the case. Wood (1999: 110–11) notes that in north China ‘proto-celadons’ were produced and that these were not of a particularly notable quality; the developmental focus was more on developing the technology of whitewares, blackwares and lead-glazed sancai wares (lead-glazed earthenwares). The kiln and workshop evidence is discussed here because it is clearly antecedent to a discussion of the evidence for celadon production.
Four areas were excavated at Yaozhou, two most extensively: area I measuring approximately 81 × 28 m and area II approximately 42 × 22 m. A series of pottery workshops and kilns were discovered in both areas. In one workshop – cave dwelling 3 in area II – a remarkable range of evidence was discovered: the shaft hole for a potters wheel towards the southern end of a workshop and at the northern end a deposit of lion figurines and bowls near a deposit of ‘clay paste’; about 2 m east of the clay deposit was a pottery jar with a groove running away from it, both presumably were used in the preparation of clay (Institute of Archaeology 1992: 16). The workshop in cave dwelling 5 in area II contained lumps of clay, another potters wheel shaft hole, more clay paste, a sagger and a deposit of unfired lamps. In cave dwelling 6, area II, evidence for the manufacture of specific pot forms was found: a deposit of spouts for handled ewers, as well as a deposit of ewers themselves. In a further workshop (Z12, area I), an area of trampled clay was discovered adjacent to pits full of clay and pottery vats, which presumably contained water. In yet another workshop (Z20, area I) a tub about 2 m long set into the floor has been interpreted as a levigation vessel, although it might be regarded as rather small for this purpose (Institute of Archaeology 1992: 32).
The kilns that were discovered are claimed to have been used for the production of lead-glazed wares, because of their proximity to the workshops in which such pottery was (probably) produced. One kiln (Y10 in area II) measured c. 3 m × 1.5 m. It consisted of a fire box at the south-west end next to a firing chamber about 50 cm higher, and, at the northeastern end, a chimney. Two other kilns were of the same type. Kiln Y12 (in area II), though not well preserved, is smaller, about 2 m long, consisting of a double chamber with a chimney at one end, the fire box presumably missing (Institute of Archaeology 1992: 40). Although there is no direct evidence for it, perhaps this smaller furnace was used for fritting the glaze components. One of the best preserved kilns (Y28, area IV) was of a similar form to Y10 and a square shape measuring about 4.5 m × 5.5 m. A series of pillars, found to be still in situ in the firing chamber, would have supported pots; the back wall of the kiln was especially well preserved providing evidence for the way in which gases and smoke was drawn out of the kiln. A pair of vents on both sides of the wall fed into the bases of the chimneys (Institute of Archaeology 1992: 41); the internal length of the largest chimney was 1.5 m so the draw on the gases must have been significant. The kiln must have been an updraft type in which the hot air and gases emanating from the fire reverberated off the roof of the kiln down onto the pottery (which was presumably located on pillars), the entire structure being of a square shape and possibly encased in a layer of clay to retain the heat within. Depending on how the pots were stacked within the kiln, the roof may have been relatively low. Having passed over the pots being fired, the gases would have been drawn down to the bottom of the back wall, through the vents and up the chimneys.
It is clear that evidence for almost all the different processes of pottery production have been found, possible levigation, trampled clay, the balls of clay which have presumably been physically prepared, the potters wheel, unfired pots, saggers for firing the pots in and the kilns with pedestals. No evidence for kiln bars was mentioned and the primary evidence for the manufacture of lead glazes appears to be lacking. The kiln form, with the fire box at one end of the firing chamber at a lower level, rather than underneath it, may be linked to a kiln tradition which included climbing (dragon) kilns with the fire box at one end.
Seven areas belonging to the Five Dynasties were excavated at Yaozhou, two of which produced most evidence: area IV measuring about 343 × 383 m and area VI measuring about 466 × 57 m (Institute of Archaeology 1997: 6).
As for the Tang kilns, the workshops belonging to this period also display an interesting range of evidence for the production of pottery, in this case celadons. In workshop Z20 in area IV the position for potters wheels, a vat in which water would have been stored, and deposits of (unspecified) clay paste were found. It is claimed that in workshop Z66, area IV an area was discovered where the clay was kneaded and a ‘washing’ tub was found – perhaps for the potters themselves. In Z17 a ‘precipitation’ (i.e. levigation) tub was found.
Five Dynasties kilns have much in common with the Tang examples described above, with a fire box lower than the firing chamber at one end and two square chimneys at the opposite end. As with the Tang kilns, different features are preserved to different extents. Kiln Y15 in area IV measures about 5 m × 4.5 m, is an enclosed horseshoe shape and is therefore of a similar size to (some) Tang kilns. This kiln, together with kiln Y29 in area VI, had a blocked entrance to the fire box. This might either indicate that the kiln was sealed during its last firing, to starve it of air, or that this was the (normal) way of producing a reducing atmosphere. Excavation of a boat-shaped kiln (Y31, area VI) revealed a line of twelve funnel-shaped saggers still in situ at the front of the firing chamber just behind the wall at the back of the fire box (Figure 4.14). This kiln also had remains of what might be the base of vents which would have transmitted the gases upwards towards the roof of the kiln. The area just behind the fire box was evidently sufficiently hot to fire pots in the saggers. A corridor-shaped entrance to the fire box (a stoke hole), would have contained the heat effectively. The arrangement of vents into the what was probably a single chimney in kiln Y43, area VI, was apparently far more complex than the other kilns: five vents along the base of the wall with other smaller square vents higher up, including seven in a second register, and although poorly preserved, more in a third register about 1.2 m above the kiln floor (see Figure 4.30). One affect of this arrangement of flues would be to draw the gases over the pots at a variety of levels, the highest being at 1.2 m. One possible inference is that the pots may have been stacked somewhat higher than in kilns with vents into the chimneys at lower levels.
Figure 4.30 Kiln Y43 a Five Dynasties kiln at Yaozhou.
Clearly it cannot be assumed that the Tang Dynasty production technology for lead-glazed wares was the same as for northern Song celadon, although, if lead-glazed wares were being produced, the kiln designs are basically the same. Considering the very much lower maturation temperatures needed for Tang Dynasty lead glazes this might be considered surprising. There is, however, a likely explanation: that the pottery bodies themselves were fired at high temperatures, the glaze applied by dipping the pot, and the glazed pot fired at lower temperatures in saggers in the same kilns – perhaps using fuel which burnt at lower temperatures.
Six areas were excavated: area II of 82 × 20 m, area III approximately 90 × 14 m, area IV, approximately 70 × 66 m, and area VI, a series of four slip trenches with a minimum width of 2.5 m wide × 60 m long (Institute of Archaeology 1998). The evidence for the production of northern Song celadon is on a massive scale, including workshops where glaze was probably applied to the pottery, and kilns with more complex designs and/or better preservation than those which date to the Shang Dynasty or Five Dynasties. There is so much excavated evidence that only a proportion will be discussed here, where it serves to illustrate differences and builds on the evidence described above.
At various points around the inside of the workshop walls were post holes, presumably for supporting a wooden roof. Workshops Z1–1 and Z1–2 in area I were constructed next to kiln Yl. Although the kiln is basically of the same type as ascribed to Tang pottery production the two workshops have new and fascinating pieces of evidence for celadon production. Workshop Z1–1 contained three large glaze vats’, the largest measuring 85 cm across which was set into the floor (see Figure 4.31). Adjacent to the vats was a ‘heatable brick bed’ with a stoke hole beneath it. Without being sure about the identity (chemical composition) of the glaze, it is difficult to suggest what function precisely the ‘glaze vats’ had. If there was no way of heating the glaze, it must have been removed in raw chunks for subsequent re-use. It is difficult to know what the ‘heatable brick bed’ was used for. The long-section of the ‘heatable brick bed’ is not drawn to scale (Institute of Archaeology 1998: 18). In Workshop Z1–2 the balance of equipment that was necessary to produce celadon was discovered (Figure 4.32): a tub which was considered to have been used for kneading the clay, a potters wheel pit, with a deep narrow hole into the ground which would have been used for holding the ‘spindle’ of the wheel in place, a stone mortar for grinding raw materials, with a water vat next to it. The turntable itself was found at the north-western end of the workshop, on a stone platform next to a deposit of saggers (Institute of Archaeology 1998: 20). A similar heatable brick bed to that in Z1–1 was found in workshop Z11 and a large one measuring 1.4 × 1.9 m in workshop Z22. Another long narrow workshop measuring 7.5 × 1.5 m (workshop Z37) contained the evidence for a similarly wide range of production activities: a deposit of ‘body paste’, a tub for kneading the clay, two pits for potters wheels, and a pottery vat, presumably to contain the water needed by the potters in building pots on the wheels. At the northeastern end of the workshop was a brick and stone platform. A pit containing fine sand was found in workshop Z42 (Institute of Archaeology 1998: 42).
Figure 4.31 Workshop Z1–1 a northern Song Dynasty kiln site at Yaozhou.
Figure 4.32 Workshop Z1–2 a northern Song Dynasty kiln site at Yaozhou.
Yet another workshop (Z45) contained the elusive evidence for the production of saggers, the vessels which protect (especially) glazed pottery from variations and excesses of kiln conditions. Unfired saggers were found near both a kneading tub and a potters wheel. The scientific investigation of some saggers was carried out (see Section 4.6.5.4 ), but not of these unfired examples. In workshop Z71 two stone seats were found near two wheels which were interpreted as potters’ seats. More evidence for levigation of clay was found in workshop Z78 in the form of two large ‘tubs’ measuring 4.4 × 3.6 m and 4.7 × 4.1 m (maximum) respectively.
Eighteen northern Song kilns were found. Most are of the same design and lay-out as Tang and Five Dynasties kilns, basically divided into three functional areas: an entrance leading into a firing pit at the lowest level, a stepped firing chamber in front of this and, at the far end, two chimneys connected to the firing chamber by vents. Where they survive, in some cases there appear to be more than one register of vents connecting the firing chamber to the chimneys as found in Tang kilns.
The survival of especially complex ‘air-inlet’ (i.e. forced draught) systems among the northern Song kilns illustrates added sophistication when compared with this aspect of Five Dynasties kiln technology. The surviving length of the ventilation duct for kiln Y56 is 3.26 m, which serves a kiln of 3.6 m in length. The (complete) and most complex ventilation duct for kiln Y63 is 3.3 m long (i.e. nearly the same length as for Y56) serving a kiln of 4.4 m in length. It is about 35 cm deep and 35 cm wide at its narrowest point, and was presumably capped with flat stones, as seen in other ventilation duct constructions. It has a bifurcating mouth widening to about 70 cm presumably into which bellows and tuyères were inserted and perhaps driven with water power. The opposite end of the duct widens to 1.07 m and has a series of three dividing walls forcing the draught through three flues over a 1.5 m deep ash pit. A similar arrangement was also found in kiln Y21, but with a duct that split into five flues, with the draught being swept over an ash pit into the kiln. Yet another (incomplete) arrangement for the air inlet system was found in kiln Y36. The evidence from the illustrations suggests that the air was forced into two separate ducts at either side of the kiln over the leading edge of the ash pit – presumably helping the fuel lying on the grate to burn. In this case the kiln was of a stepped construction. At the ‘end’ of the air inlet system in kiln Y47 is a wall across the duct which has a series of six holes near the top in order to channel the air (at higher pressure) into the fire box; each hole is only 10 cm square.
A structure, measuring about 2.3 × 2 m, was found attached to the end of the ventilation duct for kiln Y2. Although no details are given of what was found inside the box which was, according to the drawn section, only about 40 cm deep, this appears to have been the location of the bellows system. This ventilation duct was capped with flat slabs. The constructional details of the brick-built ash pit for kiln Y2 are also helpful in reconstructing what others probably looked like: the 1 m deep pit was brick built with an opening at the base for raking out the ash. Another well-preserved grate over an ash pit was found in kiln Y4. Perhaps the most complete kiln is Y5: here the full arrangement of the grate lying over an ash pit and the way that the ventilation duct fed the air into the firing chamber can be seen. The duct fed air so that when it hit the far wall of the ash pit it would have been forced upwards through the burning fuel and also downwards, presumably compacting ash in the ash pit which was lined with a vitreous ‘sintered’ layer of slag. A door in the external wall of the kiln allowed fuel to be inserted; the door lay over the below-ground air duct (see Figure 4.33). Evidence for the use of saggers, which were normally removed after use, were found as imprints in the floor of kiln Y19, towards the back of the kiln and funnel-shaped saggers were found at the front of kiln Y31 (Figure 4.14).
Figure 4.33 Kiln Y5 a northern Song Dynasty kiln site at Yaozhou.
The evidence for the fuel used is slim, except in northern Song kiln Y3, where a deposit of coal cinders was found at the bottom of the ash pit and adjacent to kiln 44 where an area in which coal was heaped was identified – and nearby a dump of ‘furnace slag’ deposited against the outer wall of the kiln (Figure 4.34). Li Guozhen et al. (1992) claim that wood was used as a source of fuel in the Tang Dynasty and Five Dynasties. The use of coal as a fuel source may have had an affect on the kiln atmosphere with the introduction of sulphurous fumes which could have led to the development of yellowish iron-sulphur chromophore in some of the glazes and the use of saggers.
Figure 4.34 Kiln Y44 a northern Song Dynasty kiln site at Yaozhou.
As mentioned above, saggers have been found in situ in the kilns and also imprints of the saggers were found in the floor of kilns. Two types of saggers have been found: cylindrical and ‘ware’-shaped (You Enpu 1986: 282, Figures 1.2). The cylindrical saggers have perforated holes in the side walls in order to allow water vapour and other gases to escape. Either single pots or stacked pots could be placed inside the saggers. The ‘ware’-shaped northern Song saggers have their rims turned downwards with a groove underneath – also used by Ding potters and very similar to those used today. When pots were placed in saggers, the potters always used a spacer of some sort, called setters. During Tang production a tripod was used to separate pots inside cylindrical saggers, as is clear from the three pin marks found on the pot bases. In the Song dynasty either one or two vessels were placed in a sagger. One kiln (Y28) had the remains of supporting posts presumably so that the gases could interact with pot and sagger comprehensively. The posts had various diameters and were either hollow or solid.
During the Tang, Five Dynasties, Song, Jin and Yuan dynasties moulds were used to produce intricate decoration on wares produced at Yaozhou. Some 200 complete moulds or mould fragments were found during the excavations at Yaozhou, of which the majority were used for the production of bowls and plates. More than eighty Tang dynasty moulds were found which were for producing, among other forms, human figures, animals, bottles, handleless cups and bell pillows; more than 90% of these were for determining the complex object shapes. They were mainly two-part moulds and some had the potter’s surname or full name carved on them. Sixteen-and-a-half Five Dynasties moulds were found at Yaozhou. Some of the moulds formed a set, having been produced from the same original. The highest quality moulds are those which date to the Song Dynasty, particularly the middle Song Dynasty: more than 90% were used for stamping decoration and many of the others were copied from the original mould (son moulds) for the mass-production of bowls, plates, cups and pillows. The impressed technique is claimed as a technical advance which was introduced in the middle Song dynasty when robust forms were decorated with fluid and elegant incised decoration (Guo Yanyi et al. 1995: 320). Compared with Song moulds, Jin moulds were heavy and the patterns less complicated. The patterns on the Jin moulds included fish, lotus and peony flowers and peacocks. Fewer Yuan moulds were found and the main pattern was peony flowers.
Guo Yanyi et al. (1995) have studied the characteristics of these moulds and identified three basic types: (1) the original mould; from which (2) the remoulding mould is copied – also known as a son mould; and (3) a body mould which is taken from an existing decorated pot body. The function of the moulds was also designated: (1) the ‘moulding’ mould, for forming complicated body shapes; (2) moulds for stamping a pattern on the pot – often the same mould as (1); (3) the ‘dressing’ mould to correct/perfect the shape of a damp pottery body.