Professional Documents
Culture Documents
4 Costin y Hagstrum 1995
4 Costin y Hagstrum 1995
Stable URL:
http://links.jstor.org/sici?sici=0002-7316%28199510%2960%3A4%3C619%3ASLISAT%3E2.0.CO%3B2-E
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/sam.html.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.
http://www.jstor.org
Fri Nov 2 18:52:06 2007
STANDARDIZATION, LABOR INVESTMENT, SKILL, AND THE
ORGANIZATION OF CERAMIC PRODUCTION IN LATE PREHISPANIC
HIGHLAND PERU
Specializution encompasses many wuys to orgunize craft production, rungingfiom .small, household-bused w,ork units to large
w ~ r k s h ~Distinctive
.~. types o f specializat~ondevelop in response to various sociul, economic, and environmentul,fuctors,
including the demand for crafts, the social relutions ofproducers, und the support base ,fir urtisuns. These factors in turn
influence man~lfacturingtechnologl: Thu.s, diff'erent types ofspecialization can he charucterized h), a "technologiculpr~?file,"
which r<flects relative labor investment. skill, und standardization. An unulysis qfPrehi.spanic cerumic technology in the cen-
tral .sierra of Peru demonstrates how these technologicul profiles can he used to identxv the wa),.s ceramic pnjduction w,us
orgunized to provision consumers with utilitarian and luxury potter?, As w,e demonstrate in our unulysis ofpottery recovered
In the Yanamarca Valley, utilitarian Wanka-st),lecooLw~uresund storage jar.s were produced by independent household-bused
artisans. while imperial Inka-st).le jars were produced by lo call)^ recruited corvie lubor w,orking.for the state.
La especializacicin economicu incluye mu1tiple.s estrategius puru orgunizar la produccicin de bienes. Estus estrategias ahar-
cun desde peyuen'as unidudes domksticas, hasta talleres de produccicin. Dijerentes y hien d~finidas,formasde especializacion
se desarrollan en re.spuesta u,fuctores .sociale.s, economicox y ambientules como la demandu por 10s bienes producidos, 1a.s
relaciones sociules establecidas enpe 10,s productores y lax buses socioeconomica.s de soporte de lox urtesanos. Esto.s fac-
tows, a su vez, infllrencian lu tecnologiu de produccion. Difewntes tipos de e.specializacion economicu pueden estar cara-
terizados por '3erfiles tecnologicos" yue rej7<jun la inversion relativa de truhujo, lu hahilidad de 1o.s productores, y lu
regulurizacicin de 10s productos. El undlisis del uspecto tecnologico de una serie de alfures prehi.spanicos tardios de lu sierru
central del Peru demuestru como 1o.s "perfiles tecnoldgicos" pueden .ser usados para identificar 1as)jrma.s en yue la pro-
duccicin de ceramica estuvo organizudu puru proveer ceramica utilitaria ), .suntuuriu a 10s consumidores. El caso ylre .se anal-
iza en detalle es el de lox Wunku de lu slerra central del Peru. En esta sociedud, ollas y otras formas utilitarias, asi como
grande.~ jurras de almacenamiento, ,fuer(jn pr(jducidus en el estilo local Wanka por urtesanos independientes c~ryonivel de
produccicin no sobrepusci la lrnidad domkstica de produccicjn. Por otro lado, lu produccion de "ariba1o.s" en el estilo Inku
imperial estuvo a cargo de trubujadores tributurlos reclutados localmente pero yue trubajuron directumente puru el estado
In ka .
nalyses of technological attributes such materials are lacking in the assemblage under
.
Cathy L. Costin. Department of Anthropology, California State University, Northridge CA 91330
Melissa B. Hagstrum Department of Anthropology, University of Washington, Seattle WA 98195
era1 trends in the organization of production. distributors) must move some distance to transfer
Such analysis is most useful in comparing differ- goods from producer to user.
ent wares from a single assemblage in terms of The third parameter is the constitution of the
their relative organization of production or production unit. It describes the group size and
change in a single ware over time. In this paper, social relations of those individuals who regularly
we study three ceramic wares from the Upper cooperate to produce a recognized corpus of goods.
Mantaro Valley to demonstrate how the organi- At one extreme is household production. Crafts are
zation of production can be investigated by tech- manufactured within a domestic setting by a single
nological analyses of manufacture. family member or a small group of related individ-
uals who reside together. At the opposite extreme is
The Organization of Craft Production the factory, or workshop in a nonindustrial setting.
Specialization is not a unitary form of economic 'This is a large facility staffed by unrelated individ-
organization, but rather must be viewed as a uals who are recruited voluntarily through wage
broad concept encompassing several distinctive contract or involuntarily through bondage, slavery,
types of organization. Costin (1986, 1991) has or mandatory labor tax obligations.
identified four parameters that can be used to The fourth and final parameter is the intensit?,
describe the organization of production. The first of production. It describes the relative amount of
parameter is the context of production, which time individual producers devote to craft produc-
reflects the nature of the demand for a particular tion relative to other economic tasks. At one
good. When referring to context, we distinguish extreme are part-time producers, individuals
between "attached" and "independent" special- whose economic strategies combine production
ization (Costin 1986; Hagstrum 1985, 1986; see of a specific craft with service, agricultural or
also Brumfiel and Earle 1987; Earle 198 1 ; Gero wage labor, or who rotate among several different
1983; Russell 1988). Attached specialists pro- crafts, working at each part-time. At the opposite
duce politically and socially symbolic goods and extreme are individuals who hypothetically
wealth that circulate primarily within the politi- devote all their economic energies to a single
cal economy and serve to maintain political craft, or even a single task, procuring all other
power and to enforce social distinctions. necessities from others in exchange for the goods
Distribution of these goods is controlled since they produce or the wages they earn.
artisans produce upon command for elite spon- These four parameters-context, concentra-
sors and patrons. In contrast, independent spe- tion, constitution. and intensity-reflect the
cialists manufacture utilitarian and domestic underlying causes of the organization of produc-
wares that circulate within the subsistence econ- tion. For example, knowing the context of pro-
omy and serve in household maintenance. duction-attached or independent-allows us to
Because there are no implicit or explicit sanc- focus on either political or economic explanations
tions on the acquisition of these goods, producers for the rise of specialization in a particular case
and consumers are free to make distribution and (cf. Brumfiel and Earle 1987). The relative con-
procurement arrangements within a framework centration of production-nucleated or dis-
of established social reciprocity or market persed-is in part dependent on resource
exchange (Hagstrum 1996). distribution, and identification of this parameter
The second parameter is the concentration of value allows us to evaluate the appropriateness of
production. It describes the spatial relationship applying resource access/allocation models (such
between producers and consumers. At one as those of Arnold [I9751 and Rice [1981]) to a
extreme, producers are uniformly dispersed particular case. The choice between part-time and
throughout the consuming population, minimiz- full-time production is an economic one. although
ing transportation time and costs from place of the considerations are different among attached
manufacture to place of use. At the opposite and independent specialists (Costin 199 1 ;
extreme, all producers are nucleated in a single D'Altroy and Earle 1985; Hagstrum 1989; Hicks
production location, and goods and consumers (or 1987; Murra 1980). Although each of these para-
Costin and Hagstrum] CERAMIC PRODUCTION IN LATE PREHlSPANlC HIGHLAND PERU 621
meters is described heuristically as a dichotomous under what conditions, and for what purposes.
variable, having two opposite values, they are in Manufacturing technology, analyzed here by
reality a continuum, with the values for the pro- labor and skill invested and the standardization
duction of any one commodity falling somewhere attained in finished goods, is a sensitive indicator
between the two most extreme values. of the sociology and economy of craft production
Combining the four organizational parameters, and consumption.
Costin ( 1 986, 1991) has proposed eight idealized
types of specialization, each of which is expected Labor Investment
to evolve under particular social, economic, polit- Labor investment refers to manufacturing costs,
ical, and environmental circumstances. Briefly measured by the time required to produce some
described they are: commodity. The relative investment of labor and
Individual specialization: autonomous individ- the degree of specialization may be interrelated
uals or households dispersed uniformly among insofar as specialized industries are competitive,
the population and producing for unrestricted emphasizing efficiency in production (Arnold
local consumption 1987; Feinman et al. 1984; Hagstrum 1985, 1986;
Dispersed workshop: larger workshops dis- Torrence 1986). An example of competitive, effi-
persed among the population producing for unre- cient production is provided by utility wares.
stricted local consumption These are the vessels produced for general con-
Communit]c specialization: autonomous indi- sumption whose manufacture reflects the con-
vidual or household-based production units, straints of a competitive economic situation. Such
aggregated within a single community, producing pottery is recognized by simple shapes lacking
for unrestricted regional consumption surface decoration.
Nucleated workshops: larger workshops aggre- Competition and efficiency, however, are not
gated within a single community, producing for always hallmarks of specialized craft production.
unrestricted regional consumption The social functions of pottery vessels affect the
Dispersed c o w i e : part-time labor producing labor invested in their manufacture. Decoration,
for elite or government institutions within a requiring more labor to execute, is positively cor-
household or local community setting related with vessel visibility because decorative
Nucleated cowee: part-time labor recruited by elements can communicate important social
a government institution, working in a special- information, for example, group affiliation and
purpose, elite, or administered setting or facility socioeconomic status (DeBoer and Moore 1982;
Individual retainers: individual artisans, usu- Thompson 1958). Craft goods produced for elite
ally working full-time, producing for elite patrons patrons encode more social and political infor-
or government institutions within an elite (e.g., a mation (Clark and Parry 1990:293-294; Earle
"palace") or administered setting 1982; Pollock 1983), requiring greater energy in
Retainer workshop: large-scale operation with manufacture, than utility items produced for gen-
full-time artisans working for an elite patron or eral consumption (Clark and Parry 1990; Costin
government institution within a segregated highly 199 1 ; Hagstrum 1986, 1989). Moreover, crafts-
specialized setting or facility people working under elite sponsorship are freed
from the economic constraints of a competitive
Demand, Technology, and the Organization market because the disposal of their wares is
of Production guaranteed.
The different ways craft producers are organized The lack of competition, the importance of
and the different demands consumers have for intelligible communication, and the practice of
crafts influence the technology of craft manufac- conspicuous consumption underlie the often high
ture. The relationships among technology, the labor intensity of attached craft production,
organization of production, and the nature of whereas efficiency in production and competition
product demand center on the contexts of manu- in distribution underlies the minimal labor invest-
facture and use: who makes pottery for whom, ment of independent specialization.
622 AMERICAN ANTIQUITY [Vol. 60, No. 4 , 1995
+ LABOR INVESTMENT -
Attached - Independent
-
Workshop - + MECHANICAL STANDARDIZATION
.Individual
-
Nucleated -+ MECHANICAL STANDARDIZATION
.Dispersed
+ SKILL -
Full-time Part-time
Figure 1. Chart illustrating the relationship between the value of the technological characteristics of craft goods and
the four parameters that describe the organization of production.
vessels that take longer to make require more implies facility and assuredness in accomplishing
hands, total output levels being equal. Third is the the production tasks. The confidence of skilled
intensity of production: all other variables being artisans, as manifest in their crafts, applies to prac-
equal, fewer full-time potters will be required ticed household potters and to highly trained work-
than part-time potters. Fourth is the length of time shop potters alike. The vagaries of quantifying skill
the assemblages represent: longer occupations by lead us to suggest that ceramic technological attrib-
definition will have been served by greater num- utes related to skill and control, to standardization,
bers of artisans, and therefore long periods or and to labor investment be considered together,
phases will exhibit greater variability than will since they combine in distinctive ways, discrimi-
shorter periods with similar forms of production. nating different kinds of specialization.
Skill Teclznological Characteristics and the
Skill reflects the craftperson's experience, profi- Organization of'Production
ciency, and talent and is recognized and appreci- Identifying the parameters of production and the
ated by artisans and consumers alike. Although types of specialization is more conclusive when
archaeologists often mention skill as an attribute the actual production locations have been identi-
of specialized manufacture, it is rarely quanti- fied (Costin 1991). Yet it is still possible to use
fied. Indeed, skill is a difficult behavioral trait to the technological characteristics of the objects
measure in ethnographic, much less archaeologi- themselves to delimit the organization of produc-
cal, contexts. Analytically, however, we may tion (Arnold and Nieves 1992; Benco 1986;
expect to recognize skill under certain condi- Hagstrum 1985, 1986; Rice 1981; Sinopoli
tions. The repetition and experience characteris- 1988). Figure 1 illustrates the relationships
tic of specialist production should lead to among the technological attributes of the prod-
regularity and consistency in technique, with ucts themselves and the parameters that describe
fewer errors as measured by manufacturing the organization of production.
rejects (Clark and Bryant 1986). The mastery of The context of production-identifying the
technologically or artistically complex produc- sociopolitical status of consumers and the nature
tion sequences is also evidence of skill. Finally, of the producer-consumer relationship-is most
skill is expected to be posititiely correlated with clearly reflected in the amount of labor invest-
the intensity of production, because artisans who ment. The products of most attached specialists
spend more time at their craft accomplish their will be more labor intensive than the products of
tasks with increasing deftness through repetition independent specialists because of the greater
and experience. information load they carry and because of the
The recognition of an artisan's technical skill desire of elite patrons to add value or uniqueness
involves a measure of subjectivity, since skill to these goods through greater labor input.
AMERICAN ANTIQUITY [Vol. 60, No. 4, 1995
The relative geographic concentration of pro- turing operations, the products of part-time work-
duction is reflected in the amount of mechanical, ers should reflect less proficiency, resulting from
and to a lesser extent intentional, standardization. interruptions and hiatuses in their craft produc-
The products of dispersed specialists should tion activities. Those industries that require
exhibit more variability than nucleated specialists greater skill and training to master technological
because workers aggregated within a single com- complexity or design elaboration will employ
munity are more likely to share raw material full-time laborers, as it is more efficient to train
sources, to participate in ad hoc sharing of facili- relatively fewer workers.
ties, to exchange tools and labor, and to have gen- Integrating the parameters of production, as
eral access to one anothers' products. they define the eight types of specialization, with
The constitution of production units' personnel the technological characteristics, it is possible to
is also reflected in the amount of mechanical stan- identify a unique "technological profile" for each
dardization. The products of large workshops are type of specialization (Table 1). For example,
likely to be more standardized than those of indi- community specialization is defined by indepen-
vidual forms of production because workshop dent context, nucleated concentration, household
artisans generally work in close proximity under constitution, and usually part-time intensity. Thus,
supervision, share technology, and draw from a it should be characterized technologically by rela-
common store of tools and raw materials. Among tive low labor investment, low to moderate stan-
attached specialists, intentional standardization dardization, and moderate skill.
should reflect the constitution of production units:
small andlor dispersed labor groups will be used The Study Sample
when standardization is unimportant (e.g., the The ceramic collections analyzed in this study
serviceable goods of dispersed corvee laborers) or were made as part of the Upper Mantaro
when uniqueness is intended (e.g.. the elaborate, Archaeological Research Project (UMARP) (see
one-of-a-kind goods produced by individual D'Altroy and Hastorf 1996; Earle et. a1 1987).
retainers), while larger workshops and factory- UMARP conducted fieldwork in highland Peru
like settings will be employed when patrons (Figure 2), investigating the social and political
require large numbers of standardized goods with changes that occurred after the indigenous Wanka
standard values (cf. Wattenmaker 1991). population was conquered by the Inka empire in
Finally, the intensity of production can be mea- roughly A.D. 1460.
sured in part by the skill manifested in the assem- Domestic contexts from four sites occupied
blage. Whereas the products of full-time after the Inka conquest of the Wanka yielded the
specialists should be more skillfully made, ceramic assemblages discussed here. The basic
reflecting continuous attention to craft manufac- unit of analysis was the patio group, an architec-
Costin and Hagstrum] CERAMIC PRODUCTION IN LATE PREHlSPANlC HIGHLAND PERU 625
Figure 2. Map of the research area. Sites indicated are (2) Hatunmarca (Wanka I1 center, Wanka I11 town); (5) Hatun
Xauxa (Inka provincial capital); (7) Tunanmarca (Wanka I1 center); (41) Umpamalca (Wanka I1 town); (54) Marca
(Wanka I11 town); (59) Wankas de la Cruz (Wanka 111 village); (74) Chucchus (Wanka I11 village). After Earle et al.
1987:Figure 1.
tural complex consisting of one or more circular slip and the Wanka decorated wares were manu-
stone structures opening to a walled patio space. factured both before and after the Inka conquest.
Each patio group probably accommodated a sin- However, only materials from Inka period (Wanka
gle household. A combination of random and 111) contexts were included in this analysis. Each
judgmental procedures was used to select 14 patio of these wares served distinct functions, and each
groups and areas within them for excavation (see met a particular demand, as characterized by level
Earle et al. 1987:7-16 for a complete discussion). of consumption, the social makeup of the con-
Both elite and commoner households-distin- suming population, and the political and symbolic
guished by size, masonry quality, and location- nature of the vessels.
were included in the sample.
Base Clara and Wanka Red
The three ceramic wares analyzed are the local
Wanka transport, storage and serving wares, These two decorated types sharedmany character-
called Base Clara and Wanka Red; the ubiquitous istics (Figure 3). Both were produced in slipped
cooking ware, called Micaceous Self-slip; and the and unslipped varieties, and were painted with
prestigious Inka-style state ware. Micaceous Self- rapidly executed black and red designs (Figure 3a,
AMERICAN ANTIQUITY [Vol. 60, No. 4 , 1995
F ~ g u r e3. Local ~ t ! l e storageitransport jars: Base Clara bod! fragment (a); \\anha Red and Base Clara bod! jar rlm
and neck fragments (b): and decorated Base Clara bod! fragments (c). ifter C ostin 1986:Figures 2.25,2.51,2.54,2.55,
2.57 and Hagstrum 1989:Figure 6.4a.
c). Morphological similarities suggest they ser\,ed nity specialization. Petrographic a~ialysessupport
similar functions. Petrographic analysis indicates this identification of a single source for the pro-
they were made of similar clays (C'ostin 1986). duction of these pottery types (Costin 1986,
Although each ware was made in s e ~ e r a ldifferent 1996).
vessel forms, only high-necked Jars were included
in our analys~s(Figure 3b). These jars were likely
multifunctional; the smaller jars used for trans- The globular lo\\-necked.iars characteristic of this
port~ngliquids and the larger ones used for storing undecorated type (Figure 3 ) were manufactured
wet and dry foods. Jars were reco~eredin both from a distinctive, highly nucaceous clay similar
elite and commoner contexts. suggesting they met to that still used today to produce cooking o1lu.v in
a broad demand. the Mantaro Valley (Hagstrum 1989). Many
Our understanding of the organization of pro- sherds had carbonized deposits on interior or
duction of Base Clara and Wanka Red is based on exterior surfaces. suggesting use in cooking.
the analysis of direct evidence for their nianufac- Cooking Lessels \\,ere reco~ered in all liause-
ture. The concentration of wasters from their holds. but \\,ere more concentrated in elite hause-
manufacture at a single town in domestic contexts holds. We suggest this reflects greater periodic
unassociated with special purpose or elite archi- demand for cooking \.essels during intermittent
tecture argues strongly for independent conimu- feasting sponsored by local elites (Costin 1986).
Costin and Hagstrum] CERAMIC PRODUCTION I N LATE PREHlSPANlC HIGHLAND PERU 627
Figure 5. lnka aryballoid jars: miniature jar form showing globular bod) and conical base (a) (after Costin
1986:Figure 2.28; Hagstrum 1989:Figure 5.32); jar rim and neck fragments (b) (after Costin 1986:Figure 2.29); and
decorated body fragments (c) (after Costin 1986:Figure 2.62).
organization of production of the Base Clara and to develop a set of hypotheses and test implica-
Wanka Red wares as community specialization. tions for the organization of production of each
Less is known about the organization of produc- ware. Specifically, we ask whether the organiza-
tion of Micaceous Self-slip and Inka pottery. tion of production of Micaceous Self-slip was
Given the relationships among the parameters of community specialization or nucleated workshop
the organization of production and labor invest- production; we ask whether the production of
ment, standardization, and skill as set forth in the Inka-style pottery was nucleated corvee or
model above, we suggest that by comparing the retainer workshops.
technological characteristics (labor investment,
standardization, and skill) of Inka, Micaceous Micaceous Selfislip: Communitj
Self-slip, and the painted Wanka types, we should Specialization or Nucleated Workshop
be able to identify the organization of production Production?
of the former two wares more conclusively. Given the general, unrestricted nature of the
Drawing on earlier studies of ceramic production demand for cooking vessels and the petrographic
in the Upper Mantaro (Costin 1986; D'Altroy homogeneity of the wares, we initially suggested
1981; D'Altroy and Bishop 1990; Hagstrum that this ware would have been produced by inde-
1986, 1989; LeVine 1987), we focus our expecta- pendent, nucleated craftspeople, either commu-
tions for Micaceous Self-slip and Inka production nity specialists or nucleated workshops. The two
Costin and Hagstrum] CERAMIC PRODUCTION IN LATE PREHISPANIC HIGHLAND PERU
lnka
Nucleated corvie moderate-high high low-moderate moderate
Retainer workshop high high high high
types of production can be distinguished by the be tested for Micaceous Self-slip, we expected
technological profiles of the manufactured that the possible types of production organization
wares. As outlined in Table 2, the assemblage proposed for Inka wares would be distinguishable
produced by community specialists is expected by their technological profiles (Table 2). An
to manifest low labor investment and moderate assemblage produced by corvte labor will mani-
skill. The amount of intentional standardization fest moderate to high amounts of labor invest-
should reflect social expectations and the range ment. Intentional standardization will also be
of vessel functions. In the case of Micaceous high, because of the need to systematize the mes-
Self-slip, such standardization should be rela- sage embodied in the wares. Mechanical stan-
tively high, since this undecorated ware served a dardization will be relatively low, reflecting the
single function, cooking. Although the amount of relatively large number of part-time laborers.
mechanical standardization should reflect the Skill is expected to be moderate: adequate for
number of work groups, we expected standard- producing the relatively elaborate or complex
ization of this kind to be low because community products, but not high because of the part-time
specialization is characterized by relatively many nature of the work.
part-time artisans. Retainer workshop production would be distin-
If Micaceous Self-slip were produced by work- guished by high levels of all the technological
shops, we would expect the assemblage to exhibit attributes. Labor investment and intentional stan-
greater standardization than one produced by dardization would be high for the same reasons
community specialists because of greater sharing they are high for nucleated corvee. In contrast
of raw and prepared materials, tools, and other with part-time corvee, mechanical standardiza-
technology by workshop members. Workshop tion is high in full-time retainer workshops,
products should exhibit less energy expenditure because these workshops employ relatively fewer
than those of individual specialists, as the relative specialists. Skill is expected to be high because
"mass production" of independent workshops the full-time nature of the work is assumed to lead
tends toward greater efficiency. to greater perfection of technique.
Inka: Nucleated CorvPe or Retainer Procedure
Workshops?
We analyzed jar fragments from all three wares,
Given that this ware was distributed through state selecting specific sherds with a nonproportional,
channels primarily to state institutions and local multistage cluster, random sampling procedure
elites, we expected this ware to have been pro- (Costin 1986:54-57; Hagstrum 1989: 144-145).
duced under some form of attached specializa- Sherds were included only if they met certain spe-
tion. The documented petrographic homogeneity cific size (greater than 4 cm2 surface area) and
and the impression of stylistic homogeneity nar- surface preservation (uneroded) criteria. A range
rowed our choices to some form of nucleated pro- of stylistic, morphological, and technological
duction, either nucleated corvee or retainer attributes was recorded for each selected sherd
workshops. As with the types of specialization to (see Costin 1986:Appendix A; Hagstrum
AMERICAN ANTIQUITY [Vol. 60, No. 4 , 1995
Table 3. Production Task Index Po~nts spond directly to the time required to execute
these tasks. Based on her experience as a practic-
Task Points ing potter and her ethnographic fieldwork among
Primary formation (points for each side) I traditional Mantaro Valley potters, Hagstrum
Ekident-<oiled 1
L
large ollas, small to large chatas, and small to ethnographic assemblages. This is likely for sev-
large tostaderus) and three shape categories and eral reasons. First, potters with limited mass-pro-
five size classes of transportistorage jars (individ- duction technology will produce more variable
ual to medium porvngos; individual to extra-large assemblages than those using standardized tools
ulpus; and small and medium tinajas). We did not or mass-production technology. More important,
recognize the range of morphological or size vari- archaeological assemblages are created over a
ability in the archaeological assemblage that Lvas much longer period than are those studied by
encountered in modern households. This may be a ethnographers. Logically, more potters should
result of the difficulty in reconstructing vessel have created the archaeological assemblages than
size and shape from an assemblage of sherds. the ethnographic ones, and therefore the former
Longacre et al. (1988) calculated coefficients should be expected to manifest greater variability
of variation on the same order of magnitude as than the later. In many ways, comparing the sta-
ours for an assemblage from Grasshopper Pueblo. tistics of an archaeological assemblage with those
The Grasshopper data, like ours, rendered coeffi- of an ethnographic assemblage to judge the
cients of variation higher than those calculated on degree of specialization counters some of the
ethnographic samples of known production orga- most important principles of such comparative
nization. Longacre et al. (1988) ultimately con- analysis, most notably that the assemblages com-
cluded that their archaeological data contained pared must have been produced over a similar
several ceramic classes unrecognized by the amount of time and must represent total popula-
archaeologists (see Stark 199 1). tions of similar size (Costin 199 1 ).
It is intriguing to note, however, that the coef- Technological Attributes. The second set of
ficients of variation on our archaeological assem- attributes analyzed consisted of paste color, slip
blage are within the same order of magnitude as color, and paint color. Ceramic color is affected
those calculated on other hand-built assemblages by several factors, the most important being the
(Crown 1991; Longacre et al. 1988), although composition of the raw materials and the firing
coefficients of variation calculated for wheel- process. Variability in color, to the extent it relates
made pottery from Late Roman and Early Islamic to the technologies of resource preparation and
sites were much smaller (Benco 1986: Figures 6 firing, should reflect the organization of produc-
and 7 ) .These studies suggest there may in fact be tion in two ways. First, we assume the variability
greater variability within hand-built archaeologi- in composition (or number of "recipes" used to
cal ceramic assemblages than in biological popu- formulate the clay bodies, slips, and paints) to be
lations, ancient mass-produced assemblages, or correlated with the number of work units to the
Costin a n d Hagstrum] CERAMIC PRODUCTION IN LATE PREHISPANIC HIGHLAND PERU
Paint Slip
Paste Red Black Red Cream
Micaceous Self-slip 1.34 - - - -
(N=552)
Wanka waresa 1.19 1.08 .90 1.02 .91
(N=94 1) (N=172) (N=256) (N=57) (N=581)
lnka 1.05 .93 .97 1.02 1.13
(N=400) (N=119) (N=133) (N= 1 16) (N=108)
Note: Shannon-Weaver diversity score (H') calculated: If' = Z p, log p,
a Base Clara and Wanka Red analyzed together for paste and black paint, separately for red paint and s l ~ p .
extent that each work unit will have its own mechanical standardization as measured by color
sources and methods for procuring and preparing variability and morphological standardization.
raw materials. Second, we assume that relative These data. then, suggest that the producers of all
uniformity in firing characteristics reflects the wares were nucleated to a similar degree, since
number of work groups because we expect each they exhibit equivalent homogeneity (or hetero-
work group to have its own consistent firing regi- geneity). This conclusion is supported by the pet-
men, including pot stacking arrangements, fuels rographic data, which indicate a single "source"
used, and cues for determining the end of firing. for each of the wares. Furthermore, the data sug-
Five color attributes-paste color, cream slip, gest similar numbers of potters or workshops
red slip, red paint, and black paint-were were at work producing each of the three wares.
recorded using standard Munsell notation of
value, hue, and chroma (Munsell 1975). Skill
Variability in Munsell values for each attribute Skill can be measured in a number of ways.
was assessed using the Shannon-Weaver diversity Standardization itself is sometimes considered a
index (Pielou 1966a, 1966b; Shannon and Weaver hallmark of potters' skill. However, this measure is
1963), calculated as useful only in industries in which the products were
H'= C pi log pi intended to be homogeneous. For those products
This statistic was developed in information whose uniqueness is desired (a hallmark of many
theory to measure the amount of noise or ran- elite wares), such a measure would be inappropri-
domness in a message. To the extent that we can ate. Although none of the industries considered in
assume the Wanka and the Inka had particular our study represents highly individual and unique
colors in mind when they decorated their vessels, pottery produced exclusively for elite patrons, we
we can assume that the variability in achieving do recognize Inka potters to have been more skilled
those colors was in fact unintentional "noise." (or at least more consistent) than those potters pro-
Although some precision in assessing variability ducing the Wanka pottery analyzed here.
is lost by treating the ordinal (or continuous) Two technical aspects of pottery manufac-
Munsell readings as nominal data, this may have ture-wall thickness and firing core-were ana-
been the most effective method for measuring lyzed to inform us of the relative skill of the
variability in multidimensional data. artisans. Pots constructed in the Wanka and Inka
The results of the analysis of color diversity are ceramic traditions were probably manufactured
somewhat ambiguous (Table 5), because none of by coiling on a muyuchiku or slow wheel
the wares was consistently more or less variable (Hagstrum 1989:161-1 62). By measuring wall
than the others. Given the lack of consistent pat- thickness on opposite sides of a potsherd oriented
terning in the color data, we conclude similar according to its axis of revolution, we expect to
amounts of color diversity in the Wanka and Inka see evidence for skill in vessel forming technique
assemblages. (including coiling, scraping, and subsequent sur-
Conclusions drawn from conzparative stan- face modification).
dardization. In sum, the three wares showed sur- For each sherd, wall thickness was measured in
prisingly little difference in their degrees of two locations. The data presented in Table 6 show
AMERICAN ANTIQUITY [Vol. 60, No. 4, 1995
Rim
Inka (N=7h) X4 .I1 0 015
M~caceousSelf-slip (N=177) 73 ,IT 0 I'll
\Vanha (N=l36) X1 II i - 088
Collar
Inha (N=17) .IX 87 0 046
Micaceous Self-sl~p(N-69) .Ih hl 25 0 1x7
Wanka (N=39) .23 4I ih 0 I'll
Body-beneath-handlea
Inka (N=4C) .I2 44
Micaceous Self-blip (N-77) l il 30
\?Bnka (Nr3X) .03 .15
V ( ~ t eVariation
. in wall thickness calculated as
n thichiir.\\ _
Maximum wall thickness ~ n ~ n i n i u i\\all
-
Var_.wt -
the variation around the mean wall thickness. cal- culated on the distributions indicates the ditTer-
culated ences to be statist~callysignificant at the .OO 1 level.
Although the quantitati\,e measures indicate
only slight skill differences among the Inka and
Summary statistics and coefficients of varia- Wanka wares. there appears to be a qualitatiw dif-
tion for these measurements are inappropriate. ference. Inka vessels appear to have been more
because the distribution more closely approxi- carefully constructed and decorated than were the
mates a Poisson distribution than a normal distrib- Wanka \,essels. indicating greater skill and:or
ution. The general pattern, howe\,er, indicates that effort on the part of the artisans producing them
the control of the Inka potters in forming their pots (compare Figures 3 and 5 ) . The complex mor-
is somewhat greater than the potters producing the phology oi'the Inks aryballoid jar form, including
Wanka and Micaceous Self-slip wares. All Inka conical base. sharp-angled carination at the junc-
wssels (100 percent) showed less than fi\,e per- ture of the base and body, and the tall smoothly
cent variation in rim and collar wall thickness. In c u r ~ e dnecks. required an understanding of the
percentage terms. Micaceous Self-slip potters drying properties of clay and careful timing for
showed less control in two of three locations than building the constituent parts of this \,essel shape.
did artisans producing Wanka-style transport6stor- hallmarks of accomplished and skilled artisans
age jars. However, this is in part a function of the (Figure 5 a ) . The qualitatiw difference recognized
differences in mean wall thickness. Because in the care and control. if not skill. to execute the
Micaceous Self-slip cooking vessels are thin painted decoration of Inha jars by contrast to
walled even a slight variation will translate into a Wanka jars is graphically illustrated by compar-
relati\,ely large percentage difference. ing Figures 3a and 5a. The Inka repertoire o f g e o -
The second variable used to evaluate the skill ot' metric design e l e ~ n e n t s (Figure 5 c ) exhibits
the potters was the presence or absence of a firing technical mastery not apparent in the Wanka
core, considered a rough indicator of the level of repertoire (Figure 3c). whic11 consists of haphaz-
control and consistency in firing. As Table 7 ard squiggles. dots. and randomly intersecting
demonstrates, Inka and Micaceous Self-slip potters lines. ljagstrum ( 1989:254) lnore often ga\,e addi-
showed significantly more controlLconsistency in tional points for "noteworthy skill'' to Inka sherds
firing than did the makers of Wanka-style jars. than to local pieces, so this "qualitative" differ-
Specifically, only I I percent of Micaceous Self- ence was quantified to some extent in the analysis
slip sherds and 17 percent of Inka sherds retained a ot' labor investment.
grey organic core, while 37 percent of Wanka (!/ .skill. 111sum, the
fi.otll i~ticlll..si.s
C'oiic~1u.sioii.s
sherds had a gray core. A chi-square statistic cal- similarities in skill level suggest that all potters
Costin and Hagstrum] CERAMIC PRODUCTION I N LATE PREHlSPANlC HIGHLAND PERU 635
producing the wares recovered in the Yanamarca Table 7. Presence or Absence of Firing Core
worked at the same relative intensity, that is, the
s a n e degree of part- or full-time work. % Absent % Present
Inka (N=404) 83 17
Summary and Conclusions Micaceous Self-slip (N=561) 89 11
Wanka (N=953) 63 37
Labor investment, standardization, and skill chi-square = 146.995 df = 2 prob < ,001
assessed in the manufacture of Micaceous Self-
slip cooking vessels, Base Clara and Wanka Red morphological standardization, related to its pri-
storageltransport jars, and Inka aryballoids are mary function, cooking. Inka manifested only
summarized in Figure 7. Inka was the most labor slightly more mechanical standardization than
intensive of the wares, followed by the Wanka either of the Wanka wares.
decorated wares and then Micaceous Self-slip. As In sum, the primary difference among the three
expected labor investment in manufacture varies wares is the amount of labor invested in manufac-
according to the social and political functions ture, indicating differences in production organi-
these wares served. Micaceous Self-slip was a zation specifically in the parameter of the context
purely utilitarian cooking ware limited to use of production. Specifically, we corroborate previ-
within domestic contexts. Base Clara and Wanka ous analyses suggesting Inka ceramics were pro-
Red vessels served multiple functions, including duced by attached specialists, while Micaceous
transportation and storage of wet and dry com- Self-slip, Base Clara, and Wanka Red were pro-
modities; these pots were used and seen outside a duced by independent specialists.
family context. Their decoration apparently sig- We argue that the similarities in standardization
naled some form of social affiliation, perhaps and skill among all the types in this study indicate
political allegiance or moiety membership they were produced by similar numbers of potters
(LeBlanc 198 1 ). Inka aryballoids-the most labor or workshops, with similar degrees of concentra-
intensive-served the most overt political func- tion, all working with the same intensity.
tion, and clearly had the most public role of any of Overall, the local storage jars and Micaceous
the wares analyzed. These highly standardized Self-slip ollas have similar technological profiles.
vessels were used throughout the empire to trans- Micaceous Self-slip did not have the relatively
port, store, and prepare symbolically charged higher degree of mechanical standardization we
co~nmoditiessuch as maize and chicha (maize would expect if the ware had been produced
beer), and were often distributed to local popula- within large, supervised workshops. We conclude
tions as part of imperial largess (Morris 1978). that Micaceous Self-slip, like Base Clara and
In Figure 7, we see that Inka involved the Wanka Red was produced by community special-
greatest amount of skill in manufacture, followed ists, independent household-based producers
by Micaceous Self-slip and Base ClaraIWanka aggregated at a few communities (Table 8).
Red. The Inka jars showed a much higher degree The Inka data are in line with expectations for
of intentional stylistic standardization, a function a supervised form of attached specialization
of their use as mass emblems of political alle- (Table 9). This is the most labor intensive pottery
giance and favor. The Micaceous Self-slip cook- consumed by the Wanka population. Those vari-
ware showed the greatest amount of intentional ables reflecting skill and control-control over fir-
Figure 7.Three primary wares ranked by technological characteristics used to identify the organization of production.
AMERICAN ANTIQUITY [Vol. 60, No. 4, 1995
ing atmosphere and variability in wall thickness- tion. We conclude that all the local Wanka wares
indicate more care in manufacture than was recog- analyzed were the products of specialists aggre-
nized for the local types. The Inka ceramics are gated within a single community and working
also intentionally standardized in terms of decora- independent of elite or bureaucratic control, a
tion. However, the Inka wares did not exhibit the type of specialization called community special-
degree of mechanical standardization that would ization. In contrast, Inka-style ceramics were pro-
indicate production by a small number of carefully duced by nucleated corvee labor, possibly local
regulated work groups working in retainer work- potters mobilized part time to produce pottery in
shops. Indeed, those variables that we contend the Inka style as their mit 'alabor obligation.
measure the number of hands working in an indus-
try, secondary morphological and technological iicknoitledgrnents. Authorship is listed alphabetically, rather
attributes, indicate similar numbers of workers than hierarchically. recognizing this collaboration as a prod-
producing Inka wares as those producing other uct of the long-term development of our two careers. An ear-
lier version of this paper was presented in the symposium,
local types. With the exception of labor invest-
"Organization and Technology o f Prehispanic Ceramic
ment, the technological profile of Inka is surpris- Production in the Andes," at the 56th Annual Meeting of the
ingly similar to those of the other wares. Society for American Archaeology, 1991
We suggest that the Inka wares recovered in The data analyzed in the paper were collected with the
Wanka households were not produced in full-time financial support of several granting agencies. including the
National Science Foundation (BNS 82-0723); the Joint
retainer workshops but rather by relatively larger
Committee on Latin America and the Caribbean of the Social
numbers of corvee laborers working only part-time Science Research Counc~l and the American Council of
throughout the year in service to the state. This is Learned Societies. with funds provided by the Ford
precisely the type of labor organization that has Foundation, National Endowment for the Human~ties,and the
been identified from written records for the pro- Andrew W. Mellon Foundation; Fulbright-Hayes: and the
Friends of Archaeology. Graduate Division, and Department
duction of Inka-style pottery at the administrative
of Anthropology of the University of California, Los Angeles.
center of Huanuco Pampa (LeVine 1987) and We would like to thank all the members of the UhlARP
among the Lupaqa of the southern altiplano (Julien who were responsible for initial field collection of data and
1982). Levine (1987:24) argues from the ethnohis- the processing of the ceramic materials. John Hildebrand and
toric documents that the potters conscripted to pro- Bruce Owen assisted with computer analyses. Luis Jaime
Castillo Butters translated the abstract into Spanish. Earlier
duce Inka wares were the same artisans who
versions of the paper were read by Philip Arnold Nancy
produced utilitarian wares for local domestic con- Benco, Tim Earle, hl~chelleHegmon, and William Longacre:
sumption. Moore (1958:56) also reports that local along with three anonymous reviewers and Michael Graves,
potters were required to produce ceramics as their they provided useful and provocative comments and sugges-
tax contribution. If the local potters who produced tions. Any flaws in the analysis and interpretation are the
responsibility of the authors.
cooking ollas and local style storage and transport
jars were recruited to produce Inka vessels, then References Cited
the slight differences in "skill" between the local
Arnold D.
and state wares might more correctly reflect super- 1975 Ceramic Ecology of the Ayacucho Basin, Peru: Implications
vision, and to a lesser extent training. for Prehistory. Currenr Anthropoiog?, 16:18>205.
Our study demonstrates how measures of labor Arnold, D., and A. Nieves
1992 Factors Affecting Ceramic Standardizatlon. In Ceramic
investment, standardization. and skill can be used Production and Distribution, edited by G. Bey and C.
to identify the organization of ceramic produc- Pool, pp. 93-1 13. Westview Press. Boulder. Colorado.
Costin and Hagstrum] CERAMIC PRODUCTION IN LATE PREHlSPANlC HIGHLAND PERU