Orton C., Tyers P., Vince A. Pottery in archaeology

Подождите немного. Документ загружается.

Firing

% 10.5, An early second-century Roman kiln from Highs»* Wood in nor,K London.

(Photo: Bernard Brandham)

130 Making pottery

also employed by potters who use the wheel (e.g. Miller 1985) - we need

not always seek kilns when faced with wheel-thrown pottery.

Somewhat intermediate between open and kiln firings are structures such

as that described by Rye and Evans from Musazi in Pakistan (Rye and

Evans 1976, 34-5). Here the firing area (3m x 10m) is enclosed by a wall

two metres high on three sides. The pots (large jars) are surrounded by

donkey dung for firing. The photographs of this structure being unloaded

(Rye and Evans 1976, plate 26c) show remarkably little sooting or burning

of the walls, or ash or wasters in the firing pit. Once more the identification

of such a structure as part of a pottery manufacturing workshop might be

difficult.

Kiln firing

Domestic ovens, such as bread ovens, may be pressed into service to fire

pottery (Desbat 1989, 147) but it is more usual to build a specialised kiln

structure (fig. 10.5).

The study of pottery kilns has generated a considerable body of litera-

ture, both ethnographic and archaeological, and it is only possible here to

point to the more significant elements. The classic studies by Hampe and

Winter of modern pottery and tile kilns in Crete and Cyprus (Hampe and

Winter 1962) and southern Italy, Sicily and Greece (Hampe and Winter

1965), with their excellent photographs, provide a wealth of valuable infor-

mation on kiln and workshop structure and many other aspects of pottery

manufacturing. They have proved to be the starting point for many others

considering kiln morphology (for example kiln sections republished by

Peacock 1982, fig. 12, 3-8 and McCarthy and Brooks 1988, fig. 21). Gazet-

teers by Vossen and others of modern potteries in Spain (Vossen 1972;

Vossen et al. 1980) and Morocco (Vossen and Ebert 1976) maintain this

tradition. Kiln structure and operation is also considered in the surveys of

pottery production in Pakistan (Rye and Evans 1972) and India (Saraswati

and Behura 1966).

Turning to the archaeological record, classifications have been compiled

for Roman kilns in Britain (Swan 1984), Gaul (Duhamel 1973) and Italy

(Cuomo di Caprio 1971-2) and Peacock (1982, 67-70) has discussed the

possible relationships between them. For medieval kilns in Britain we have

the studies of Musty (1974; see also McCarthy and Brooks 1988, 40-54).

Thus anyone facing the prospect of excavating a pottery kiln has a bewil-

dering quantity of comparative material to consider, but will hopefully avoid

the pitfalls of the rather eccentric kiln reconstructions that have occasionally

appeared in print. Swan (1984, 128) emphasises the importance of

sufficient

and detailed recording of the kiln structure itself and recommends in par-

ticular a continuous section drawn across the structure and its infill from

front to back, a section through the kiln structure at right angles to the flue

Firing

131

axis and careful consideration of the stratigraphy in the flue. All fragments of

kiln furniture and baked clay from the vicinity of the kiln should also be

examined and recorded. Far too little is known about the superstructure of

most excavated kilns to allow any potential information to slip by

unrecognised.

POTTERY FABRICS

We here consider the role of the examination of fabric in pottery studies.

Historically, form and decoration have received more attention from archae-

ologists than the details of the fired clay itself. Since the 1960s a more

systematic approach to fabric analysis has come to take one of the central

positions in pottery studies. It was seen as one of the means of breaking out of

the strait-jacket of primarily chronological concerns and expanding the scope

of the study into the areas of technology, trade and exchange. These develop-

ments cannot be divorced from other shifts in archaeological thinking and

practice over the same period.

The methods of fabric analysis have been largely drawn from the geological

sciences, for the reasons succinctly given by Peacock: 'Pottery can be

regarded as a metamorphosed sedimentary rock and thus it can be argued

that ceramics are best approached in a manner similar to that used in the

geological study of the parent raw materials' (Peacock 1977, 26). The tech-

niques have had their impact not only in the university research laboratory,

but all the way down to the pot-sorter grouping and describing sherds in the

finds hut.

Why look at fabrics?

The physical characteristics of the fired clay, its appearance and composition,

are determined by:

(i) the natural composition of the raw material(s);

(ii) the actions taken by the potter in creating the clay mix, such as

settling out the coarser component, adding non-plastics, or com-

bining two or more clays;

(iii) the firing atmosphere and temperature;

(iv) the use and post-depositional environment of the vessels.

Most of the stages of the manufacturing process (the workability of the

clay, the size, shape, wall thickness and decoration of the formed vessel, the

shrinkage rate during drying and possible firing techniques) and the technolo-

gical properties of the finished product (its porosity, physical and thermal

strength) are all to a great extent dependent on the character of the original

clay mix, notably the frequency, size, shape and identity of the non-plastic

inclusions.

132

Why look at fabrics

133

The examination of pottery fabrics - the composition and structure of the

fired clay body - gives valuable information on three broad themes: (1) the

technology of the manufacturing process; (2) the physical characteristics of

the fired product; and (3) its provenance.

Manufacturing technology

There are a variety of techniques that can be used to determine the maximum

temperature reached during the firing cycle (Rice 1987, 426-7, table 14.3).

Some require relatively simple equipment such as an electric kiln or a

penological microscope, whereas other techniques are confined to those with

access to more complex facilities. Many of the techniques rely on the identifi-

cation of changes in the clay body which take place at high temperatures,

such as the changes in the porosity or level of vitrification of a

fabric.

Some of

these changes rely not only on the maximum temperature achieved during the

firing cycle, but also the time that the temperature is held and other variables

such as the firing atmosphere. Thus maximum firing temperature is only

being measured indirectly. The determination of firing temperature is impor-

tant for two aspects of the pottery. Firstly there is the relationship between

firing temperature and firing technology. A series of pots which are fired to a

narrow range of temperatures may suggest a reasonably sophisticated level of

control over the firing process. Secondly, a series of vessels consistently fired

to a temperature exceeding the point of vitrification may be intended to be

non-porous, and this may be an important indication of their original

function.

The firing atmosphere has a significant effect on the fabric of a fired vessel,

most notably on its colour. The fired colour of a fabric depends principally on

the iron compounds and carbon that the clay contains and the duration,

temperature and atmosphere during firing. Most clays, but in particular

sedimentary clays, contain a proportion of organic matter. When heated in

an oxidising environment (where there is an excess of

free

oxygen) the carbon

will burn and form carbon dioxide, which will attempt to escape from the

fabric. In some cases the escaping gas causes bloating or bubbling. Where

there is no excess of oxygen (a reducing environment), or the duration of the

firing is insufficient, the carbon will not all burn out, but will be visible in the

final fired fabric as a dark grey or black core. This core may be particularly

marked where the vessel is thick, such as at the rim or base, or in the section

of the handle. However, carbon can also be deposited on the surface of the

pot during firing in a process known as 'smudging'. This results in a very

black surface and margins.

Iron compounds in the clay also react to the firing atmosphere. In oxidising

conditions the iron compounds in the clay will usually be converted to ferric

oxide (Fe

), which is red in colour, but this change will not usually take

place until after the carbon has been burnt off. Even if the surplus of oxygen

134 Pottery fabrics

mwmmmmsm

Fig. 11.1. Stylised cross-sections comparing variations in the appearance of firing cores in fine

textured clays (Column A) and coarse-textured clays (Column B).

1: Oxidised, organics not originally present; no core

2: Oxidised, organics may or may not have been originally present; no core

3-4; Oxidised, organics originally present; diffuse core margins

5: Reduced, organics not originally present: diffuse core margins

6: Reduced, organics not originally present; black or grey may extend completely through the

wall leaving no 'core'

7: Reduced, organics originally present; diffuse core margins

8: Reduced, organics may or may not have been originally present; no core

9-10: Reduced, cooled rapidly in air; sharp core margin

11: Reduced, cooled rapidly in air, reduced and cooled rapidly in air again; sharp core margins,

'double core' (Rye 1981, fig. 104)

Visual examination

135

is very marginal or prolonged firing will result in a red colour, but where there

is a shortage an iron-rich clay will usually be grey (see p. 69).

The character of the atmosphere may alter several times during the firing

cycle, and as the reactions described above are reversible the final colour of

fabric may be quite complex. It is not unusual to see sherds with layers of

colours from the surface to the core, each corresponding to a different stage

of the firing and the extent to which its effect has 'penetrated' the wall of the

vessel. The appearance of the section under different firing conditions

(fig. 11.1) has been illustrated by Hodges (1962,61, fig. 1) and Rye (1981,116,

fig. 104).

Provenance studies

Provenance studies are concerned with identifying groupings, usually known

as fabrics or wares, which reflect their origin, and where possible determining

their source. The 'ideal' arrangement may be where a single ware represents

the output of a single workshop but this may not always be realised. Rather, a

single fabric may represent the output of a group of workshops sharing the

same raw materials and processing them in a similar manner, and conversely,

a single workshop (or related group of workshops) may be responsible for

more than one fabric (because of the use of different sources of clay process-

ing systems). Recognising that this may be the case does not negate the

procedure of fabric characterisation (as occasionally seems to be suggested),

but merely adds another range of possibilities to the interpretation of the

groupings.

We recognise three stages of examination: visual, penological and com-

positional, reflecting a rising level of complexity (and cost), but decreasing

accessibility and a consequent narrowing of the quantity of material that may

be processed. The bed-rock of pottery processing procedures can only be

accurate and informed visual examination, and where possible there must be

feed-back from any results obtained from more sophisticated analyses.

Visual examination

With the realisation of the importance of consistent recording systems,

capable of reuse and cross-referencing, it is now usual for pottery fabrics to

be described using standardised categories and keywords on pro-forma sheets

or cards (see p. 233). A number of such recording schemes have been

described (Peacock 1977, 26-32; Orton 1979; Steinstra 1986; Bauer et al.

1986. Kunow et al. 1986; Schneider 1989) which follow broadly similar lines.

An initial aim of most studies is to sort the sherds in the collection into

wares, giving a single description which covers the variation within the group

rather than describing every catalogued item in detail. The visual examin-

ation should take place on a clean section through the sherd which exposes

the core. The existing breaks may not be suitable because of minerals

13 6 Po ttery fabrics

deposited on their surface during burial and it is usual to make a new clean

break. Some workers have suggested that a flat section is preferred for this

examination (Smith 1972; Glock 1975) which might necessitate the sawing of

many thousands of sherds. Sections perpendicular to the rim might, it i

suggested, also be an aid to drawing (Holladay 1976). Such a system is not

widely employed and it is more usual to make a fresh break by snipping off a

corner of the sherd with a large pair of pliers or pincers. Grains protruding

from the surface of the break appear in three dimensions, making them easier

to identify than they would be in a section and a break gives a better view

'under the surface' of the sherd. The quality of the fracture itself may be a

valuable diagnostic feature. When making the break it is important to take

account of the uneven distribution of inclusions in the fabric which might

cause the break to occur along a weaker, but perhaps unrepresentative, zone.

In wheel-thrown ware many inclusions will become aligned to the direction of

throwing, so breaks at different angles through the sherd may give rather

different views of the fabric of the pot. In the case of any doubt more than one

break should be made. A small hand-lens (x 8) or a higher power binocular

microscope (up to x 30) is invaluable as an aid to examination of the fabric

and identification of inclusions. It is important to confirm the scale of any

graticule in the eyepiece of a binocular microscope before using it for any

measurements.

Colour

Since Shepard extolled the virtues of the Munsell colour system for the

recording of pottery colour ('The advantages ... are so great that it is hardly

necessary to argue its superiority': Shepard 1956, 107) it has been widely

employed in the Americas, and since the late 1960s (e.g. Franken and

Kalsbeek 1969) in Europe.

Munsell colours are referred to by three variables, known as the hue, value

and chroma. Hue refers to the position of the colour in the spectrum. In the

Munsell system they are referred to by letters or pairs of letters:

R Red YR Yellow-Red

Y Yellow GY Green-Yellow

G Green BG Blue-Green

B Blue PB Purple-Blue

P Purple RP Red-Purple

Working around the spectrum the hues are further subdivided by a numer-

ical prefix from 0 to 10, with 0 towards the red end of the spectrum and 10

towards the purple. In the standard charts the prefixes are 2.5, 5, 7.5 and 10.

Value refers to how light or dark the colour is, with 0 representing black and

10 representing white. The chroma is the saturation or purity of the colour,

with 0 representing neutral greys and higher numbers indicating purer

Visual examination

137

colours. A Munsell colour is expressed in the order

hue value chroma,

with a

space between hue and value and an oblique stroke between value and

chroma; thus 2.5YR 6/8. As a special case, pure white, grey and blacks (that

with chroma of 0) are prefixed N (neutral).

Although the Munsell notation provides a complete description of a

particular colour it should always be accompanied by a brief

verbal

descrip-

tion. This indicates to the reader the general range of colours represented in

the material and provides a means of deciding whether resort to the colour

charts for a more precise representation of the colour is necessary. However,

neither the colour names suggested by Munsell, nor Shepard's suggested

system of modifiers (Shepard 1956, 110), have found wider favour.

The complete Munsell book of color is a large and unwieldy volume (not to

mention expensive) and for most practical purposes the smaller

Munsell soil

color

charts is preferred (Munsell Color Company 1975). The range of reds,

browns and yellows in the latter covers most of the colours encountered in

ceramic work, and the charts are contained in a durable, pocket-sized (13cm

x 19cm) volume. The standard collection of soil colour charts contains seven

hues: 10R, 2.5YR, 5YR, 7.5YR, 10YR, 2.5Y and 5Y. The sheets are perfo-

rated with holes through which both sherd and colour chip can be viewed at

the same time, and masks are supplied which should be used to reduce

interference from other colours on the chart when attempting comparisons.

Two supplementary charts are available in the same format (7.5R and 5R)

which are more red than 10R and may be

useful

for some material. Advice on

recording Munsell colours will be found in the documentation distributed

with the chart. Particular attention should be given to the recommendations

on the lighting conditions and background when the chart is used.

An alternative to the Munsell soil color chart is the Rock-color chart

produced by the Geological Society of America (1948), which employs the

Munsell notation. The chart includes selections from hues 5R, 10R, 5YR,

10YR and 5Y which partially duplicate those of the

Munsell soil color chart.

In addition there are a small number of chips in yellow, green, blue and

purple hues (5GY, 10GY, 5G, 10G, 5BG, 5B, 5PB, 5P and 5RP) which would

be useful

for glazes, slips and paints. The

Rock-color chart is

not perforated to

allow the sample and colour chip and sherd to be examined at the same time,

but it would be relatively easy to make it so,

Alternative colour systems have been used in the description of archaeo-

logical ceramics, particularly in continental Europe. Kunow and others

recommend the use of Munsell for most fabric colours but supplement this

with the Schwaneberger colour chart (designed for stamp collectors) for

coloured glazes, largely on the grounds of cost and availability of the full

Munsell system (Kunow et al. 1986, 33). In France, Germany and Switzer-

land the colour charts produced by the CEC (Fédération Européenne des

Fabricants de Carreaux Céramiques), the DIN (Deutsche Industrie Normen)

138

Pottery fabrics

standard colours (DIN 6164) or the code Cailleux (Cailleux and Taylor 1963)

are widely used (Schneider 1989, 39), although the Munsell system is gaining

popularity. It would clearly be preferable if usage standardised around a

single system, and Munsell offers the clearest advantage.

The use of the Munsell system is sometimes criticised for providing

'falsely accurate' picture of the sherd or the colour divisions are considered

'too fine'. Such remarks suggest some misunderstanding about how and when

colour charts should be used. When describing wares rather than individual

sherds it is necessary to cover the range of colours represented, but also make

it clear which are the most common variants. More seriously, the value of

colour as a means of characterising fabrics at all has been questioned (for

example Picon's dismissal of the value of 'codes de couleur compliqués':

Picon 1984). It is undoubtedly true that variations in colour, and other

factors such as porosity and hardness, may be due to firing or post-

depositional conditions rather than differences in the original materials. But

it is equally true that the slip or fabric colour produced by particular

workshops are often quite precise and regular and provide a valuable means

of distinguishing one producer from another.

If the colour is to be recorded it is better that it is with reference to some

accepted and commonly available standard rather than in vague terms or

colloquialisms in a foreign language. The reproduction of a list of oxide

percentages from compositional analysis is not an adequate substitute, is of

no value to anyone faced with the task of sorting material in the field, and in

any case ignores the effects of firing. With experience it can be recognised

when a colour is likely to be a significant element in the characterisation of a

ware and when it is not.

Hardness

Hardness is usually expressed in terms of resistance to scratching. Mohs'

10-point scale of hardness is often used but an alternative test using a

fingernail (Mohs 2 or 2.5) and a steel blade (6) has found wide favour

(Peacock 1977,30). Rice suggests that copper wire (Mohs 3) and window glass

(4.5) can be used to mark other points in the scale (Rice 1987, 357; see p. 235).

Whatever the precise items in the scale, it is rather more difficult to

determine exactly what is being measured in a composite material such as

pottery. The length and duration of firing, porosity, grain-size distribution,

post-depositional environment and mineral composition all contribute to the

'hardness' and although it should continue to be measured and recorded it

should not be used as a precise indicator.

Inclusions

Special attention should be given to the inclusions in a fabric, for in many cases

these provide the most reliable method of distinguishing between fabrics.