You are on page 1of 16

A Provisional Theory of Ceramic Abrasion Author(s): Michael B. Schiffer and James M.

Skibo Reviewed work(s): Source: American Anthropologist, New Series, Vol. 91, No. 1 (Mar., 1989), pp. 101-115 Published by: Wiley on behalf of the American Anthropological Association Stable URL: http://www.jstor.org/stable/679740 . Accessed: 21/01/2013 01:41
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley and American Anthropological Association are collaborating with JSTOR to digitize, preserve and extend access to American Anthropologist.

http://www.jstor.org

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

MICHAELB. SCHIFFER
JAMES M. SKIBO

UniversityofArizona

A Provisional Theory of Ceramic Abrasion


This article calls attentionto ceramicabrasion as a class of tracesthat can be employedby the archeologistin a varietyof behavioralstudies. Basic principles of ceramicabrasionare setforth, in the Laboratoryof Tradiundertaken built upon the engineeringliteratureand on experiments Abrasion mechanismsare discussed,but the limitations of currentknowledge tional Technology. are underscored. The majorfactors pertaining to ceramic, abrader, and contactsituations that the rate and natureof abrasionare identified. influence

MAJORTREND IN ARCHEOLOGICAL ANALYSIS during the past several decades has been the recognition and exploitation of new lines of evidence for inferring past human behavior and environmental conditions. Experimentation often plays a crucial role in validating new lines of evidence; recall, for example, the many hundreds of experiments required to establish the relationships between lithic polish types and worked materials (Keeley 1980; Vaughan 1985). This article calls attention to ceramic abrasion, a family of common traces that so far has been little studied. In order to lay a foundation for the use of these traces in inference, we have been investigating ceramic abrasion experimentally. Building upon our experimental results and recent studies in engineering, this article proposes a provisional theory of ceramic abrasion. Knowledge of the factors influencing abrasion rates and the formation of abrasion traces can contribute to three areas of archeological inquiry. First, an understanding of the technical choices that influence abrasion resistance, such as firing temperature and in specific surface treatment, can help investigators to determine if-and how-pots technological traditions were designed for good abrasion resistance. Second, abrasions or scratches comprise an abundant type of wear; thus, an appreciation for the conditions that cause abrasive wear and the factors that lead to differences in abrasion patterns can enhance ceramic use-wear analysis (cf. Bray 1982; Griffiths 1978; Hally 1983; Schiffer 1989). Third, because a host of processes from trampling to fluvial transport abrade ceramics, an experimentally grounded theory of abrasion can contribute to fine-grained identification of formation processes (cf. Schiffer 1987).

Basic Concepts and Principles


Abrasionsand Abrasive Processes Two interrelated phenomena are of interest here: (1) abrasions, as a category of traces, and (2) abrasive processes, the activities of people and nature that produce abrasions. For present purposes, an abrasion is defined broadly as a trace that was formed by removal or deformation of material on a ceramic's surface by mechanical contact, specifically, the sliding, scraping, or, in some cases, striking action of an abrader (i.e., a particle, object, or surface). The abrader can also include particles entrained in a fluid such
B. SCHIFFER is Professor,DepartmentofAnthropology, UniversityofArizona, Tucson,AZ 85721. JAMESM. SKIBO MICHAEL is a doctoralstudent in the Department of Anthropology, University of Arizona, Tucson. Both are senior staff membersof the Laboratoryof Traditional Technology,Building 30, University ofArizona, Tucson.

101

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

102

AMERICAN ANTHROPOLOGIST

[91, 1989

as water or air. Abrasions that can be observed individually tend to be linear or slightly curvilinear troughs, and archeologists usually term such features scratches, striations, abrasions, or grooves (e.g., Griffiths 1978; Hayden 1979; Shipman and Rose 1983). In many cases, however, individual abrasions are not distinguishable, especially where an area has been abraded repeatedly. An abrasive process, simply put, is one in which an abrader inflicts abrasions upon a ceramic. Moving a pot on a house floor is an abrasive process, as is the fluvial transport of sherds. This definition is problematic in that a process that creates abrasions can also produce other types of surface alteration. For example, fluvial transport and trampling lead to chipping of susceptible edges in addition to striations. Moreover, some processes can create more than one abrasive trace. Sand blasting produces both striations and microchipping, depending on the angle at which the particles impact the surface (Moore and Swanson 1983). The lack of process purity, however, does furnish the archeologist with additional traces for identifying specific processes. Rates and TracesofAbrasion Ceramics vary greatly in their susceptibility to damage by a given abrasive process. In order to compare the abrasion resistance of various ceramics as well as to quantify the gross effects of different processes, one needs a standardized measure. Percent weight loss has been used in our experiments because it is so conveniently quantified. Abrasion processes also differ markedly in the kind of damage they inflict. For example, some processes remove temper and matrix at the same rate, whereas others preferentially remove matrix. The discussions below consider both quantitative and qualitative aspects of ceramic abrasion. We have tested ceramic abrasion resistance with lapidary tumblers (see Skibo and Schiffer 1987; Vaz Pinto et al. 1987). Briquettes are weighed and placed in a tumbler barrel with an abrasive of sand, pea gravel, or gravel. After tumbling (wet or dry) for a specified time, the briquettes are reweighed, and abrasion resistance is expressed as percentage weight loss. A sample of 5-10 briquettes is employed in most tests. Although not a panacea, tumbler testing has permitted us to identify many variables that influence the rate and outcome of abrasive processes. AbrasionMechanisms The following consists of a selection of mechanisms, from the engineering literature, believed to be relevant for studying abrasion of archeological ceramics. Although adequate appreciation for some mechanisms requires high-power optical or electron microscopy, we have confined our observations to the naked eye and to low-power magnification. We also note that the mechanisms are not all independent, nor are they at the same level of abstraction. For example, several mechanisms, which can be called "primary," underlie the operation of others. Finally, closer scrutiny of abraded archeological ceramics may lead to the recognition of new mechanisms. We now briefly list these mechanisms, underscoring that at present they are poorly defined and little understood. Intergranular fracture (e.g., Mehrotra 1983; Ritter, Rosenfeld, and Jakus 1985) takes place when contact breaks the bonds between grains (clay-clay and clay-temper), thereby dislodging them. Given the great heterogeneity of grain size and composition in most archeological ceramics as well as the low degree of sintering often achieved, the bonds between grains are readily broken by many kinds of contact. We suspect that this primary mechanism figures importantly in most abrasion processes and underlies other mechanisms such as plastic deformation, plowing, and microchipping (see below). Tribochemical mechanismsaffect abrasion processes by causing chemical changes (e.g., Fischer and Tomizawa 1985). For example, water weakens intergranular bonds, and so increases the ceramic's susceptibility to material removal by abraders (Skibo and Schiffer 1987); the tribochemical effects of olive oil, brine, and other liquids also need to be investigated. These mechanisms are probably important in the depositional environment

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schifferand Skibo]

CERAMICABRASION

103

and in vessels used for storing and processing liquids. Like intergranular fracture, tribochemical mechanisms are primary and influence the operation of other abrasion mechanisms. Plastic deformation(e.g., Mehrotra 1983) occurs when contact with an abrader causes localized compression of the surface, as in the scratching of a soft ceramic with a nail. The effects of plastic deformation are attenuated in heavily tempered wares, and this mechanism probably plays little role in the abrasion of vitrified and glazed ceramics. Plowing (e.g., Wallbridge, Dowson, and Roberts 1983) involves the removal of material along a linear or curvilinear track as a result of tangential contact by an abrader. Traces of plowing, which probably depends on intergranular fracture, are commonly observed on archeological specimens. Plow scars intergrade morphologically with plastically deformed scratches or grooves. In softer (usually low-fired) wares, plowing is probably accompanied by some plastic deformation. Microchippingis the removal of small chips by (hertzian) fracture (cf. Buckley and Miyoshi 1984:337; Neema and Pandy 1977:391; Wallbridge, Dowson, and Roberts 1983:209). We believe that this mechanism contributes to the formation of plow scars, especially in relatively hard (and glazed) ceramics. In addition, edges and sharp curves in most ceramics are susceptible to microchipping by a host of abrasive processes. Flat or gently curved surfaces struck by abraders at high angles of incidence can also undergo microchipping (e.g., Moore and Swanson 1983). Fatigue and delaminationtake place in areas receiving repeated impacts (Tabor 1977). Lateral cracks form below the surface, probably at grain boundaries, and eventually the surface delaminates. This mechanism might contribute to the development of an abraded area on the bottom of a pot repeatedly set down upon-and scraped along-a rigid, abrasive surface. In archeological ceramics, fatigue (i.e., strength loss) can also be caused by thermal cycling. In particular, the heavily abraded interior of some cooking pots may result from thermally induced fatigue and delamination promoted by abrasive cleaning. Plastic flow is the rearrangement (by mechanical contact) of surface material at a subgranular level into rigid surface features. In plastic flow, as opposed to plastic deformation, boundaries between larger grains become less distinct (see Hines, Bradt, and Biggers 1977:463). Similar mechanisms, perhaps tribochemically based, may be responsible for producing use-wear polish on stone tools (Adams 1986). Plastic flow is likely to occur on higher-fired ceramics and on harder temper particles, possibly contributing to polishing under certain conditions of abrasion. Pedestalling (Skibo 1987; Skibo and Schiffer 1987) is a mechanism in which abraders preferentially remove smaller or softer grains, leaving behind larger or harder ones (i.e., temper) that protrude (Figure la). This mechanism has been described (but not named) in the engineering literature (e.g., Ayers et al. 1983; Hannink, Murray, and Marmach 1983). In order for pedestalling to occur, there must be some small abraders like sand or fine gravel that can strike the surface between temper particles. Fluvial abrasion, sand blasting, and certain uses create pedestalling. First Principles As noted above, in order for abrasion to take place there must be dynamic contact between an abrader and a ceramic. The abrasion of multiphase materials is strongly influenced by the mode of contact (and probabilities of contact) between abrasive particles and different constituents of the ceramic (e.g., clay and temper), as the following oversimplified discussion illustrates. When an abrader contacts the clay portion ofa ceramic, only two outcomes are possible: either abrasion occurs or it does not. When an abrader impacts a temper particle, however, there are generally three possible outcomes: the temper particle may be (1) abraded, (2) removed (dislodged), or (3) neither abraded nor removed. Whether abrasion occurs or not (actually, the extent of abrasion) depends upon the hardness of the abrader in relation to that of the clay or temper. The removal probability of a temper particle is influenced by (1) the extent to which it is already pedes-

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

104

AMERICAN ANTHROPOLOGIST

[91, 1989

Figure 1 The effects of firing temperature on the abrasion resistance of briquettes tempered with fine sand: (a) 550'C, (b) 6500 C, (c) 750' C, and (d) 850' C. (Length of pictured area is ca. 3 cm.) Note extreme pedestalling on (a). tailed, (2) the strength of the clay-temper bond, and (3) the momentum and contact angle of the abrader. The outcome of the abrasive process is also affected by the likelihood that abraders will contact clay or temper particles. These probabilities are in turn determined by (1) the density and distribution of temper and (2) the size of the abrader in relation to temper size. Rates of abrasion differ from place to place on a ceramic's surface, depending on local topography. Holding constant all other factors, rates of abrasion vary directly with surface convexity. Thus, marked convexities like edges and cracks experience higher abrasion rates because the probability is greater there that any contact will dislodge clay or temper particles (cf. Desai et al. 1983; Larsen-Basse 1985).

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schiffer and Skibo]

CERAMICABRASION

105

It is convenient to present more fine-grained principles of abrasion in relation to (1) characteristics of the ceramic, (2) characteristics of the abrader, and (3) nature of the ceramic-abrader contact.

Characteristics of the Ceramic


Strengthof Fired Paste Strength (or hardness) of the fired paste, defined here as resistance to intergranular fracture (cf. Rice 1987:356), has a profound effect on the rate and course of abrasion. Strength is determined by variables that the potter could manipulate during the manufacture process, including clay composition and firing temperature. In addition, strength is degraded by some uses, including cooking and water storage, as well as by environmental processes such as freeze-thaw and wet-dry cycles. Clay composition influences ceramic strength, but the magnitude of such effects is presently unknown. Different clay minerals (e.g., illite, kaolinite) begin to melt at varying temperatures and may form chemical bonds with other constituents of the paste that differ in nature and strength. Alkalis and iron and calcium compounds can act as fluxes during firing; that is, by combining with certain clay minerals they form a glassy phase that imparts greater strength to the fired product (Grimshaw 1971:274-275, 278, 280). In short, there are ample grounds for expecting clay composition to influence abrasion resistance. Of the variables that the potter can control (including temper, see below), firing temperature has the greatest effect on abrasion resistance. Firing temperature (and duration) determine the degree ofsintering and glass formation in the fired ceramic. At higher firing temperatures, clay-clay and clay-temper bonds are strengthened, thereby increasing abrasion resistance. Skibo, Schiffer, and Reid (1989) examined the relationship between firing temperature and abrasion resistance for a series of briquettes fired at 5500C, 6500C, 7500C, and 8500C and abraded in dry pea gravel (Figure 1). A marked increase in abrasion resistance was observed as firing temperature rose. In a similar experiment with an abrader of wet pea gravel, Vaz Pinto et al. (1987) showed that briquettes fired at 9000C were, depending on temper, at least 5-7 times more resistant to abrasion that those fired at 7000C. Firing atmosphere can also affect abrasion resistance under certain conditions. Grimshaw (1971:882) notes that some compounds, such as ferric oxide-a common constituent of clays-impart a greater fluxing action when fired in a reducing atmosphere. and Voids Pores, Cracks, Any factor that increases the porosity of a ceramic will decrease its abrasion resistance (see Hines, Bradt, and Biggers 1977; Champagne and Dallaire 1985). For example, clays naturally high in organic matter, which often burns out during firing, have greater porosity and a correspondingly lower abrasion resistance. Similarly, organic tempered briquettes have less abrasion resistance than untempered and even mineral tempered briquettes (Vaz Pinto et al. 1987; Skibo, Schiffer, and Reid 1989). Processes of manufacture introduce cracks and voids that, because of their effects on surface topography, decrease abrasion resistance. Cracks are also introduced after manufacture by activities of use and processes of the depositional environment. In particular, the post-firing thermal history of a ceramic should influence abrasion resistance. Each time a ceramic is heated and cooled, its constituents expand and contract differentially, gradually reducing the strength of clay-clay and clay-temper bonds. Cracks formed in thermally cycled ceramics should render them more susceptible to abrasion. Hardnessand Shape of TemperParticles The effects of temper on abrasion processes, only now being investigated (Skibo, Schiffer, and Reid 1989; Vaz Pinto et al. 1987) are highly variable: abrasion rates can be accelerated or decelerated (in relation to an identical untempered ware).

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

106

AMERICAN ANTHROPOLOGIST

[91, 1989

Temper hardness in relation to the clay matrix (and abrader) influences the operation of several mechanisms. If temper is softer than the abrader and clay, many abrasive processes will scour temper particles until their surface is depressed below the surrounding clay. The rim of such a depression, being a convexity, should itself abrade faster. When temper particles and clay matrix are of equal hardness, abraders will cause a uniform removal of the surface. For example, high-fired floor tiles in an area of heavy foot traffic show much wear, but the surface is nearly smooth with the exception of plow scars in both matrix and temper (Figure 2). As Skibo (1987) has shown, temper particles harder than the clay matrix ordinarily abrade very slowly, and are gradually pedestalled as the surrounding clay is worn away (this assumes that the abrader is at least as hard as the clay) (cf. Figure la). How far the pedestalling process goes before the particle is dislodged depends on a variety of factors discussed below. It must be stressed that only hard temper particles can be pedestalled and dislodged. There has been little study of cases in which clay or temper particles are harder than the abrader. In such situations, the abrader will wear at a faster rate, but some abrasion of the ceramic-perhaps through plastic flow-is to be expected. For example, hand contact on vessel handles can create polishes, probably through plastic flow abetted by tribochemical mechanisms. affect the course of abrasion or rounded, for example-can Temper shape-angular in several ways. (These effects pertain to hard temper particles, as shape is not apt to influence abrasion rates of particles as soft or softer than the clay matrix.) Once temper of hardness-should becomes exposed, all rough and angular particles-regardless abrade at higher rates than rounded particles because of their pronounced convexities. The effects of temper shape on dislodging are more complex. Angular particles (and those with rough surfaces) form stronger bonds with clay (Shepard 1965:27), and so should be more resistant to dislodging. However, shape irregularities could raise the probability that a given contact with an abrader would result in dislodging. For example, a ladle

Figure 2 Well-fired floor tile showing even wear of temper and matrix. Note plow scars, which indicate axis of foot traffic. (Length of pictured area is ca. 7 cm.)

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schiffer and Skibo]

CERAMICABRASION

107

moving across the bottom of a pot might be more likely to dislodge angular than rounded particles (assuming equal pedestalling). Size and Quantityof Temper If the amount of temper (by volume) is held constant, temper size can influence abrasion processes in several ways. One size effect should be present on marked convexities, such as rims, especially with hard temper. Larger particles in these situations should be less vulnerable to dislodging because of the greater area of the clay-temper bond. Because larger particles will resist dislodging for longer periods, they will themselves experience more abrasion, and may be especially susceptible to plastic flow and polishing. On flat to be separated by there are fewer of them-tend surfaces, large particles-because a surface that should be This small than areas of presents greater particles. exposed clay differentially susceptible to various kinds of abraders (Skibo and Schiffer 1987). In experiments with large and small temper (sand and crushed basalt at concentrations of 20% and 40%) Vaz Pinto et al. (1987) found that briquettes with larger temper were abraded by pea gravel at slightly higher rates. In another series of experiments, Skibo and Schiffer (1987) confirmed these findings of Vaz Pinto et al. (1987), and also discovered that ceramics with coarse sand temper had a greater abrasion resistance than those with fine-sand temper when abraded (wet or dry) with sand or large gravel. Thus, under some conditions of abrasion large temper particles can create more abrasion-resistant pastes than small ones. Organic temper presents an interesting case. Skibo, Schiffer, and Reid (1989) compared abrasion rates of briquettes containing cut grass and manure temper, the latter consisting of much finer particles. The larger grass particles had a reduced abrasion rate, apparently because their effects on surface topography were so localized. The effects of temper quantity are very much mediated by temper hardness and other factors. In the case of organic temper, more temper leads to higher rates of abrasion (Vaz Pinto et al. 1987). Different amounts of hard temper, however, have variable effects, depending on temper size and shape. Vaz Pinto et al. (1987) found that an increase in sand temper (fine and coarse) from 20% to 40% raised abrasion rates, whereas a similar increase in the concentration of basalt temper (fine and coarse) lowered them. Because of difficulties in controlling for firing temperature in these experiments, the investigators could not state whether high concentrations of basalt temper actually improved abrasion resistance over untempered ceramics. However, their results do suggest that (1) ceramics with an abundance of tightly bonded temper particles should be more resistant to abrasion than untempered ones, and (2) an abundance of weakly bonded temper can raise abrasion rates above untempered ceramics. Distribution and Orientationof TemperParticles The distribution of temper in a paste need not be uniform. For example, forming and finishing processes can depress temper particles below the surface, reducing the apparent temper density. In such cases abrasion will take place in stages: the ceramic will first behave as if it had a low temper density, and only after the surface is worn away will the full influence of the temper particles on abrasion processes be realized. During sand blasting of refractory bricks, Engman (1983) found that the rate of abrasion slowed down after the cement matrix had been removed, exposing the more resistant temper particles. Clearly, local variation in temper density will cause differences in abrasion rates. As is well known, pottery-forming techniques can affect temper orientation (e.g., Grimshaw 1971:406; Rye 1981:61). Paddle-and-anvil technology, for example, aligns temper particles with their long axes parallel to the vessel's surface. The use of rolling pins to prepare test briquettes can lead to similar alignments. The most obvious effect of this phenomenon is to cause differences in abrasion rates of surface (aligned) and interior material. For example, the greater exposure of flattened, soft temper particles near the surface should promote higher abrasion rates. Although hard temper has more variable

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

108

AMERICAN ANTHROPOLOGIST

[91, 1989

effects (see previous section), flat-lying hard temper should be more securely bonded because of the greater area of the temper-clay bond, thereby improving abrasion resistance. Although few experiments have thus far investigated the relative influence of the various temper factors on abrasion processes, some general patterns are emerging. Of the factors enumerated above, particle hardness has by far the greatest effect. The ranks of other factors are less clear, but particle shape and size as well as temper quantity are probably next in importance. Shapeand SurfaceCharacteristics of the Ceramic The shape and surface topography of a ceramic piece influence its susceptibility to many abrasion processes. As noted previously, marked convexities such as rims, handles, edges, and corners are easily abraded. As these convexities wear down, rates of abrasion gradually decrease. In experiments using wet pea gravel to simulate fluvial abrasion, Skibo (1987) showed that abrasion rates drop off markedly for freshly broken sherds as their edges become rounded. Topographic differences also lead to differences in abrasion mechanisms. For example, in our tumbling experiments microchipping develops quickly on edges and corners. At the same time, material is being removed from flat surfaces through other mechanisms, probably intergranular fracture and plowing. When cobbles are used as the abrader in tumbling tests, there is a greater intensity of microchipping as well as the removal of larger chips; flat areas show pockmarking, perhaps the result of fatigue and delamination. Concave surfaces or protected areas may not abrade at all because contact is prevented. For example, a large abrader cannot strike areas adjacent to sharply outflaring rims or under handles. These shape characteristics greatly influence the development of wear patterns on vessels. Potters employed many surface treatments, each differing in microtopography. For example, polishing or burnishing produces a relatively smooth concentration of smaller particles on the surface, causing the latter to be relatively devoid of microconvexities. In comparison, a scraped surface has a rougher microtopography. Pilot studies in our laboratory have shown that differences in surface treatment do translate into differences in abrasion resistance as predicted (cf. Rice 1987:355); compared to slipped and untreated surfaces, polished ceramics appear to have greater abrasion resistance (Baum 1985). When surfaces are not slipped or polished, temper can affect surface topography, depending on the specific finishing technique. For example, scraping a vessel in the leatherhard stage drags temper particles across the surface, thereby creating features that resemble plow scars (Figure 3). The joint effects of temper and finishing techniques on abrasion processes still need to be explored experimentally.

Characteristics of the Abrader


Ceramics are affected by a host of abraders, including sand, metal, dirty hands, and even other ceramic artifacts. The nature and amount of abrasion resulting from such contacts depend in part on characteristics of the abraders themselves, including size, mass, shape, and rigidity, which influence the amount of force that is applied at the point of contact (Avery 1977). These properties are considered one at a time, holding all others constant. Hardness The hardness of an abrader influences its effects on clay and temper. As already noted, harder abraders should cause higher rates of abrasion because more material is lost during each contact (cf. Moore and King 1980). On the other hand, softer abraders may contact a greater surface area, promoting more widespread abrasion. Although the effects of abrader hardness on mechanisms are poorly known, we hypothesize that abraders softer than clay and temper will operate mainly by plastic flow and tribochemical mecha-

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schiffer and Skibo]

ABRASION CERAMIC

109

Figure 3 Striations formed while the vessel was still leather hard. (Length of pictured area is ca. 3 cm.) nisms, whereas a fuller range of mechanisms should characterize the action of harder abrasives. Shape Abrader shape affects the amount of force that can be applied at any point. For example, a spherical abrader cannot concentrate force to the same extent as one with angular projections. Thus, in some cases abrasion rates ought to increase as abraders depart from spherical shapes (cf. Moore and Swanson 1983). Angular abraders should also produce more damage by plowing. Size Abrader size can be expressed in relation to the size of temper particles and the mean distance between them. A small abrader has a maximum diameter less than (1) the mean size of temper particles, or (2) the mean distance between them. Ordinarily, silt and sandsized particles are small abraders. A medium abrader, usually in the size range of gravel, is larger than the mean distance between temper particles but smaller than twice the maximum distance. Large particles, of course, are everything else. It is assumed in this discussion that size is a convenient measure of an abrader's mass. In general, one can expect abrasion rates to vary directly with abrader size (cf. Moore and King 1980). Experiments in our laboratory have explored this relationship (Skibo and Schiffer 1987). Briquettes of several types (untempered, fine-sand temper, and coarse-sand temper) were tumbled wet and dry with three abraders: fine sand, pea gravel, and large gravel. In every case abrasion rates varied directly with abrader size (Figure 4). For example, mean percentage weight losses of untempered briquettes after four hours of wet abrasion were 0.73 (sand), 3.63 (pea gravel), and 10.02 (large gravel). Although temper size influenced material loss (see above), these differences were dwarfed by the effects of abrader size.

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

110

AMERICAN ANTHROPOLOGIST

[91, 1989

Figure 4 The effects of abrader size on abrasion resistance and abrasion traces after dry tumbling. Briquettes have coarse sand temper and were fired at 9000 C: (a) fine sand abrader, (b) pea gravel abrader, (c) large gravel abrader. (Length of pictured area is ca. 3 cm.)

Abrader size also has important influences on abrasion mechanisms (Buckley and Miyoshi 1984). Skibo and Schiffer (1987) documented some differences in the surfaces of briquettes abraded by fine sand, pea gravel, and large gravel. Assuming relatively mild contacts, it can be proposed that smaller abraders induce little microchipping and can break only the weaker intergranular bonds. Large abraders are more likely to cause microchipping as well as fatigue and delamination. In addition, large abraders are less capable of pedestalling hard temper. In Figure 4, for example, although the large gravel abrader has removed more clay matrix than the pea gravel abrader, the latter has led to more sharply defined pedestalling. Eventually, it should be possible to infer the size (or

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schiffer and Skibo]

CERAMICABRASION

111

size mix) of abraders by examining the traces of specific mechanisms. A lack of pedestalling in low-fired wares, for example, often indicates an absence of small and medium abraders (unless the removal mechanism is mainly fatigue and delamination). Discussion The preceding treatment of abrader effects has necessarily been oversimplified. The mutual influences of the variables discussed above and their interaction with pottery characteristics and the nature of the abrasive process are very complex, and few experiments have addressed these issues. Nonetheless, the factors identified may have some value in explaining variability in abrasions observed on archeological ceramics. We anticipate that abrader size and hardness will be invoked most often because these factors probably have the greatest effects.

Nature of the Ceramic-Abrader Contact


Factors pertaining to the ceramic itself and to the abrader are insufficient to account for variability in ceramic abrasion. Ceramics and abraders can interact in different ways, and so aspects of the contact situation must also be examined for their effects on both rates and traces of abrasion. Velocity,Directionality, and Rate of Contact In order for abrasion to occur, ceramics and abraders must be in relative motion. There are three abrasionmodes: (1) the abrader moves and the ceramic is stationary (e.g., scraping during manufacture, sand blasting in deserts, and many situations of use), (2) the ceramic moves and the abrader is stationary (e.g., frost heaving and some use situations), and (3) both ceramic and abrader are in motion (e.g., certain fluvial processes and use situations). Holding constant all other factors, the greater the contact velocity, the higher the abrasion rate. Moreover, at higher velocities, each contact is apt to remove more material. These relationships should hold for most abrasion mechanisms. The direction of the abrader's motion during contact also influences the outcome of the abrasive process. (Again, it is a question of relative motion, for the abrader can be stationary while the ceramic moves.) Unidirectional motion, for example, usually produces striations (plow scars and plastic deformation) that parallel the direction of motion. For example, a large sherd from the northern Tucson basin, thought to have been used in agricultural activities, exhibits considerable abrasive rounding of one long edge. Originating from that edge are perpendicular striations that by the indicated direction of motion suggest use as a scoop or hoe (Figure 5). Striations on floor tiles can also indicate the axis of traffic flow (Figure 2). The force applied by the abrader to the ceramic can also affect abrasion rates and traces. The greater the applied force, the greater the amount of material removed. For example, a ladle used with a heavy hand will scrape more material from a vessel's wall than one used gently. Although the rate of contact directly influences the rate of abrasion, the relationship is not linear. In most abrasive processes, the removal of previous material alters the ceramic's shape and topography. For example, as sherd edges become rounded by fluvial abrasion, the abrasion rate decreases (Skibo 1987). On the other hand, we would expect abrasion rates to increase sometimes. For example, the abrasion rate of cooking pots by ladle contact might increase as protective surface treatments like polishing are worn away. It has been assumed so far that abrasion takes place when a ceramic and an abrader come into contact. This model does describe many abrasion processes, but still others involve more complex contacts. We now introduce two additional contact situations: heterogeneous abraders and abraders with a substrate. Abraders Heterogeneous Heterogeneous abraders, consisting of many particles having different properties, are common during ceramic use. For example, floor tiles receive contacts from people, ani-

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

112

AMERICAN ANTHROPOLOGIST

[91, 1989

Figure 5 Heavily abraded sherd from the northern Tucson basin. Note rounding of edges and vertical striations, especially on right half of photo. (Length of pictured area is ca. 6 cm.) mals, furniture, toys, and cleaning tools. The traces of different abraders can accumulate as patches (Schiffer 1989) in different zones of a ceramic. For example, an abrasive patch from ladling or cleaning might be found on a cooking pot's interior, whereas different patches will form on the bottom (from contact with fire dogs or floor) and sides (from cleaning and incidental contacts). In other cases, the traces of heterogeneous abraders will be superimposed alternately on the same patch. For example, the patch of wear on the bottom of a cooking pot can be caused by diverse abraders and contacts in locations of vessel storage, cleaning, and use over a fire. Heterogeneous abraders can also act sequentially, contributing to a patch that changes as it develops. This phenomenon can be expected when pots are reused. For example, the basal patch of a cooking pot reused for storage should undergo changes, depending on specific differences in abraders and contact situations.

Abraders withSubstrates
A second type of complex contact arises when there is a sandwich consisting of substrate, abrader, and ceramic. Abrasion occurs upon movement of the substrate, ceramic, or both. An example of sandwich contact is a pot being moved along an earthen floor. The floor is the substrate and individual particles, both on and in it, comprise the abrader. Another example is furnished by the abrasive wear of ceramic floor tiles: the ceramic is stationary and the substrate (human feet and various artifacts) is in motion. Abrasive particles themselves may be affixed to the substrate (e.g., sand on the bottom of shoes) or loose (e.g., "dirt" on the floor). In sandwich contact, properties of the substrate (in addition to those of the abrader and ceramic) influence the rate and course of abrasion. Relatively rigid substrates, like the sole of a shoe or a hard-packed earthen floor, ought to lead to higher rates of abrasion than more yielding substrates (such as a floor of loose sand). Moreover, abrasive mechanisms should vary. For example, very rigid substrates with attached abraders like grind-

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schiffer and Skibo]

CERAMICABRASION

113

ing wheels should produce plowing and temper dislocation but little pedestalling of temper.

andPlacement of Contact Exposure


With few exceptions, the surface of a ceramic rarely experiences uniform contacts. Thus, abrasion is concentrated in areas exposed to abraders: sand-blasted sherds erode on their upper (exposed) surface, a tile floor wears in areas of heavy traffic, and a serving vessel abrades on the bottom interior and base. As noted above, abraded regions are termed patches, and these form in a regular manner as the abrasive process proceeds (Schiffer 1989).

Effects of Liquids
Water and other liquids are readily taken into porous and permeable ceramics during use and in the depositional environment, and so influence tribochemical mechanisms. The effects of liquids on rates of abrasion have been ascertained for some engineering ceramics, and these effects (including that of water) are variable-sometimes retarding, sometimes accelerating material loss (e.g., Danyluk et al. 1983; Fischer and Tomizawa 1985; Hines, Bradt, and Biggers 1977; Wallbridge, Dowson, and Roberts 1983). Skibo and Schiffer (1987) examined the effects of water by tumbling three types of ceramic briquettes (untempered, fine-sand tempered, coarse-sand tempered) with three abraders (fine sand, pea gravel, coarse gravel); these nine experiments were carried out both wet and dry. Briquettes tumbled wet abraded at a rate 1.62 to 6.52 times faster than those tumbled dry. Ware and Rayl (1981) showed experimentally that ceramics immersed in water for long periods undergo marked strength losses. As a result, we anticipate that long-term exposure to water would create an even greater differential in wet and dry abrasion. Skibo and Schiffer (1987) also found that dry-abraded briquettes have rougher surfaces than those abraded wet, suggesting possible differences in mechanisms.

Conclusion
The framework presented above is but a first approximation to a theory of ceramic abrasion. It has identified major factors of the ceramic, abrader, and contact situation that influence the rate and nature of abrasion. In some areas, the theory has built upon extant experimental studies in engineering and in archeology. In other areas, however, our formulations are offered as hypotheses, consistent with other elements of the theoretical framework. In any event, it is clear that much more experimental work will be needed to flesh out and strengthen the theory of ceramic abrasion if we are to explain rigortraces of abrasion found on archeologichl ceramics. ously-and use in inference-the Partial support for this research was provided by grants from the National SciAcknowledgments. ence Foundation (BNS 83-10609, BNS 84-19935) to M. B. Schiffer. Gifts from Harcourt Brace Jovanovich and the Stephen Tyler Fund (of the University of Arizona Foundation) to the Laboratory of Traditional Technology contributed greatly to our effort. Jenny Adams first discovered the potential utility of the tribology literature; the senior author is indebted to her for discussions on this subject. We are grateful to Alan P. Sullivan III, Jenny Adams, Gordon Bronitsky, Michael Deal, Timothy Kaiser, Charles Redman, and an array of anonymous referees, who gave us valuable comments on drafts of this article. Paul Fish loaned us the specimen illustrated in Figure 5. J. Jefferson Reid graciously allowed the senior author to roam the Grasshopper ceramic collection, which served as a stimulus to discovery.

References Cited
Adams, Jenny L. 1986 A Use-wear Analysis Comparing Handstones Used to Process Hides and Corn. M.A. thesis, Department of Anthropology, University of Colorado.

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

114

AMERICAN ANTHROPOLOGIST

[91, 1989

Avery, Howard S. 1977 Classification and Precision of Abrasion Tests. In Wear of Materials 1977. W. A. Glaeser, K. C. Ludema, and S. K. Rhee, eds. Pp. 148-157. New York:American Society of Mechanical Engineers. Ayers, J., et al. 1983 Abrasive Wear Studies of Laser Surface-melted Aluminum and Titanium Alloys with Carbide Additions. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 265-271. New York: American Society of Mechanical Engineers. Baum, Mindy 1985 Laboratory Notebook on Experiments Concerning Surface Treatment and Abrasion Resistance. Manuscript on file, Laboratory of Traditional Technology, Department of Anthropology, University of Arizona. Bray, Alicia 1982 Mimbres Black-on-White, Melamine or Wedgewood? A Ceramic Use-wear Analysis. The Kiva 47:133-151. Buckley, Donald H., and Kazuhisa Miyoshi 1984 Friction and Wear of Ceramics. Wear 100:333-353. Champagne, B., and S. Dallaire 1985 Wear Resistance of TiB2-Fe Cermets. In Wear of Materials 1985. K. C. Ludema, ed. Pp. 33-38. New York: American Society of Mechanical Engineers. Danyluk, S., et al. 1983 Abrasion and Wear of Silicon in Fluid Environments. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 187-193. New York: American Society of Mechanical Engineers. Desai, V. M., et al. 1983 Effects of Carbide Size on the Abrasion of Co-base PM alloys. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 32-37. New York: American Society of Mechanical Engineers. Engman, U. 1983 Results from High Temperature Erosion of Ceramics. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 319-324. New York: American Society of Mechanical Engineers. Fischer, T. E., and H. Tomizawa 1985 Interaction of Tribochemistry and Microfracture in the Friction and Wear of Silicon Nitride. In Wear of Materials 1985. K. C. Ludema, ed. Pp. 22-32. New York: American Society of Mechanical Engineers. Griffiths, D. M. 1978 Use-marks on Historic Ceramics: A Preliminary Study. Historical Archaeology 12:6881. Grimshaw, Rex W. 1971 The Chemistry and Physics of Clays and Allied Ceramic Materials. 4th edition. London: Ernest Benn. Hally, David J. 1983 Use Alteration of Pottery Vessel Surfaces:An Important Source of Evidence for the Identification of Vessel Function. North American Archaeologist 4(1):3-26. Hannink, R. H.J., M. J. Murray, and M. Marmach 1983 Magnesia-partially Stabilized Zirconias (Mg-PSZ) as Wear Resistant Materials. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 181-186. New York: American Society of Mechanical Engineers. Hayden, Brian, ed. 1979 Lithic Use-Wear Analysis. New York: Academic Press. Hines, J. E., Jr., R. C. Bradt, and J. V. Biggers 1977 Grain Size and Porosity Effects on the Abrasive Wear of Alumina. In Wear of Materials 1977. W. A. Glaeser, K. C. Ludema, and S. K. Rhee, eds. Pp. 462-467. New York: American Society of Mechanical Engineers. Keeley, Lawrence H. 1980 Experimental Determination of Stone Tool Uses: A Microwear Analysis. Chicago: University of Chicago Press. Larsen-Basse,J. 1985 Binder Extrusion of Sliding Wear of WC-Co Alloys. In Wear of Materials 1985. K. C. Ludema, ed. Pp. 39-44. New York: American Society of Mechanical Engineers.

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

Schiffer and Skibo]

CERAMICABRASION

115

Mehrotra, P. K. 1983 Mechanisms of Wear in Ceramic Materials. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 194-201. New York: American Society of Mechanical Engineers. Moore, M. A., and F. S. King 1980 Abrasive Wear of Brittle Solids. Wear 60:123-140. Moore, M. A., and P. A. Swanson 1983 The Effect of Particle Shape on Abrasive Wear: A Comparison of Theory and Experiment. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 1-11. New York: American Society of Mechanical Engineers. Neema, M. L., and P. C. Pandy 1977 Erosion of Glass When Acted Upon by an Abrasive Jet. In Wear of Materials 1977. W. A. Glaeser, K. C. Ludema, and S. K. Rhee, eds. Pp. 387-391. New York: American Society of Mechanical Engineers. Rice, Prudence M. 1987 Pottery Analysis, A Sourcebook. Chicago: University of Chicago Press. Ritter,J. E., L. Rosenfeld, and K. Jakus 1985 Erosion and Strength Degradation in Alumina. In Wear of Materials 1985. K. C. Ludema, ed. Pp. 1-7. New York: American Society of Mechanical Engineers. Rye, Owen 1981 Pottery Technology. Washington: Taraxacum. Schiffer, Michael B. 1987 Formation Processes of the Archaeological Record. Albuquerque: University of New Mexico Press. 1989 A Research Design for Ceramic Use-Wear Analysis at Grasshopper Pueblo. In Pottery Technology: Ideas and Approaches. Gordon Bronitsky, ed. Boulder, CO: Westview Press. Shepard, A. O. 1965 Ceramics for the Archaeologist. Washington: Carnegie Institution of Washington, Publication 609. Shipman, Pat, and Jennie Rose 1983 Early Hominid Hunting, Butchering, and Carcass-processing Behaviors: Approaches to the Fossil Record. Journal of Anthropological Archaeology 2:57-98. Skibo,James M. 1987 Fluvial Sherd Abrasion and the Interpretation of Surface Remains on Southwestern Bajadas. North American Archaeologist 8:125-141. Skibo,James M., and Michael B. Schiffer 1987 The Effects of Water on Processes of Ceramic Abrasion. Journal of Archaeological Science 14:83-96. Skibo,James M., Michael B. Schiffer, and Kenneth C. Reid 1989 Organic Tempered Pottery: An Experimental Study. American Antiquity. (In press.) Tabor, David 1977 Wear-A Critical Synoptic view. In Wear of Materials 1977. W. A. Glaeser, K. C. Ludema, and S. K. Rhee, eds. Pp. 1-11. New York: American Society of Mechanical Engineers. Vaughan, Patrick C. 1985 Use-wear Analysis of Flaked Stone Tools. Tucson: University of Arizona Press. Vaz Pinto, Ines, et al. 1987 Effectsof Temper on Ceramic Abrasion Resistance: A Preliminary Investigation. Archeomaterials 1:119-134. Wallbridge, N., D. Dowson, and E. W. Roberts 1983 The Wear Characteristics of Sliding Pairs of High Density Polycrystalline Aluminum Oxide Under both Dry and Wet Conditions. In Wear of Materials 1983. K. C. Ludema, ed. Pp. 202-211. New York: American Society of Mechanical Engineers. Ware,John, and Sandy Rayl 1981 Laboratory Studies of Differential Preservation in Freshwater Environments. In The Final Report of the National Reservoir Inundation Study, Volume II. Pp. 3i-3-108. Santa Fe: National Park Service, Southwest Cultural Resources Center.

This content downloaded on Mon, 21 Jan 2013 01:41:51 AM All use subject to JSTOR Terms and Conditions

You might also like