Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT
Cloth was the most economically, politically, and rituaUy valuable item in the late prehispanic Andes,
and it might be expected that producers, distributors and consumers of this valuable would compete
to capture the symbolic significance of these goods as an attribute or identity for themselves. Much
has been said of the elite patrons of the textile arts in the Andes; far less has been said of the
producers. Cloth production traditionally employed domestic — largely female — labor for both utili-
tarian and wealth production. The Inka state expanded textile production by taxing conquered
households and by developing two new categories of specialist weavers. In reorganizing the economy,
the Inka redefined the social identities of cloth producers to ensure that the symbolic significance of
cloth accrued to the state, rather than to producers. The traditional category of domestic producers
were dissociated from their products, likely seeing their own status diminish as the value of their
products increased. The aqUakuna category of female specialist weavers enjoyed an idealized high
status commensurate with the value of the exquisite cloth they wove, but at a high personal cost, as
they remained sequestered virgins under control of the state.
r> emn c t
^ ^ y> a n d social status. So central was this
principle to the Inka ideology of material culture
that indigenous creation myths described the first
Cloth was the most economically, politically, people as entering the world already dressed in gar-
andrituallyvaluable item in the late prehispanic Andes, ments that identified their gender, ethnicity, and so-
In the Inka empire, cloth was a utilitarian good, me- cial (elite) status (Guaman Poma 1980[1615]:84;
dium of taxation, and conspicuous sign of wealth Santacruz Pachacuti Yamqui 1968[1613]: 284-287).
and political favor. Textiles expressed social rela- At the time the Spanish conquered the Inka empire,
tionships and social identity. Social connections were strict sumptuary laws were aimed at ensuring that a
constructed and expressed in the dynamics of tex- person's clothing would immediately signal appro-
tile distribution. As an item of exchange, cloth flowed priate rank, ethnicity, and place of residence. Con-
in many directions among individuals and institu- quered populations were required to retain their dis-
tions of different rank and power, reflecting the tinctive "traditional" local dress styles after incorpo-
nature of the social relationship between giver and ration into the empire (Cobo 1979[1653]), as a way
receiver. Textiles circulated among consanguinal and of controlling the population and maintaining eth-
affinal kin to mark adjustments to social networks nic distinctions within the empire. This was an im-
during key rites of passage such as birth, puberty, perial strategy in which only select portions of the
marriage, and death. The bestowal of cloth to so- empire were "Inkanized" after incorporation into
cial or political inferiors signified reward and pa- the empire (cf. Costin and Earle 1989) so as to dis-
tronage; the prestation of cloth to political equals or courage alliances among conquered people and de-
superiors demonstrated alliance and fealty (Murra ter rebellion. Commoners were forbidden to wear
1962, 1975 1989; Gayton 1973; Niles 1992). a range of garments and materials reserved for
The form and decoration of cloth itself — political elites, including feathers, vicuna, and metals,
apart from the social relationships it mediated -- ex- Penalties for not wearing the proper costume or
pressed components of social identity such as gen- unauthorized wearing of sumptuary articles ranged
124 Cathy Lynne Costin
from beating to death (Moore 1959:165-173). For cally, culturally, and linguistically distinct backgrounds.
provincial elites and ethnically Inka individuals, cloth- Although many works on the Inka have stressed poli-
ing indicated ones' place within the dominant cul- cies of homogenization and standardization in im-
ture. Particular kinds of cloth were a badge of of- perial rule (most recently Rowe 1982; Julien 1988),
fice and insignia of rank. After imperial conquest, recent studies suggest that imperial leaders adopted
some local elites were coopted into the Inka bu- a range of nuanced economic and political strate-
reaucracy — often dubbed "Inka-by privilege" — and gies to achieve imperial goals of pacification and
formed a new class of state administrative retainers. revenue maximization (e.g., Morris 1988; D'Altroy
These individuals were enculturated in part through 1992; Malpass 1993; D'Altroy and Hastorf 1998).
a direct subsidy of Inka-style cloth (Betanzos What is clear is that in the process of subjugating a
1987[1551]: 179; Murra 1989) and through gifts of diverse population, the Inka coopted and altered in-
women (as concubines and second wives) trained in digenous ideologies of gender relations (e.g.,
the production of this fine state-style clothing. Silverblatt 1987), reciprocity and redistribution (Mor-
Most studies of cloth in the Inka empire have ris 1982), and household/community self-sufficiency
looked at the meaning of cloth from the point of (Murra 1980).
view of consumption, and not from the point of Of particular interest has been the ways in
view of the circumstances of production. Nearly which imperial institutions produced or commanded
everyone in the empire participated to some lesser goods within the imperial political economy (e.g.,
or greater degree in cloth production, either in the D'Altroy and Earle 1985; LaLone and LaLone 1987;
procurement and processing of raw materials or in LeVine 1987; D'Altroy 1992; Spurling 1992;
spinning, weaving and finishing. However, not all Hayashida 1995; Costin 1996a). In addition to gen-
individuals or categories of persons participated eral tribute requirements enacted on the populace at
equally in cloth production in terms of labor in- large, several categories of state workers — some
vested, definition of symbolic content, or social rec- permanent, some not — produced a range of goods
ognition of their labor. Given the variety of cloth for the use of state institutions, personnel, and the
types and uses in the Inka empire, it should come as ruling elite.
no surprise that a range of social actors produced
cloth within households and for the state. Because INKA CLOTH: TYPOLOGY, VALUE
all imperial subjects participated in the "making and
meaning" of socially significant cloth — but did so
AND MEANING
in vastly different ways — it seems worth exploring The 16th century chroniclers describe several
the relationship between cloth production and so- types of cloth produced in the Inka empire. Chusi is
cial identity. a term sometimes used to denote extremely coarse
In this paper, I argue the social identities of cloth used for blankets, rugs, and carrying and stor-
weavers were not inconsequential in terms of the age sacks. Most commonly, the 16th century chroni-
values of various types of cloth or in terms of socio- clers distinguished between two types of cloth used
political relations within the empire. I argue the so- for garments, a rough cloth called 'awasqa and a finer
cial identities of weavers were related directly to the cloth called qompi. Geoffrey Spurling (1992:215-219)
meaning and relative political significance of textiles argues that this European categorization of cloth
in the Inka empire and that to fully understand the into either 'awasqa or qompi obscures a more com-
social and political roles of textiles in Inka society plex Andean system of cloth typology and nomen-
we must understand the organization and relations clature. He notes that some chroniclers used modi-
of cloth production and distribution. I will suggest fiers to distinguish among different types of 'awasqa
there was a strong correlation among the social iden- and qompi, and that early dictionaries listed a large
tities of weavers; the functions of cloth; imperial number of native terms for weaving technologies
strategies to manipulate gender, class and ethnic iden- and styles. I recognize that not all 'awasqa was of the
tity as a method of social control and engineering; same coarseness nor all qompi of the same degree of
and state strategies to control politically charged fineness, but for purposes of this discussion I main-
goods and privileged knowledge. tain the simplified categorization of cloth garments
into 'awasqa and qompi.
'Awasqa was plain weave cotton and llama-
THE INKA EMPIRE wool cloth woven from relatively coarse thread.
'Awasqa was the cloth used for the everyday clothing
The Inka Empire — Tawantinsuyu — domi- of commoners and most soldiers.
nated the western coast and Andean highlands of Qompi was decorated cloth made of fine wool
South America during the Late Horizon period (usually alpaca or vicuna, although there is one refer-
(1400-1533 C.E.). At its height, it incorporated 8- ence to a royal garment woven of bat wool [Cobo
12 million people from dozens of politically, ethni- 1979[1653]:245]) or perhaps cotton. Most qompi was
Late Prehispanic Andes 125
made of multi-colored, dyed thread woven in de- materials had symbolic connotations. For example,
signs of varying elaborateness. Some "plain" qompi gold and silver - both the metals and the colors -
was completely faced with small colored feathers were associated with the sun and the moon, them-
or metal disks that covered the cloth surface entirely selves powerful signifiers of the main Inka deities
The Andean peoples as a whole developed an in- (Lechtman 1979; Morris 1995). Design elements,
credible repertoire of techniques for making and too, were important signifiers of rank and status (see
decorating cloth (see d'Harcourt 1962) although cer- discussion below and Stone-Miller 1997). More-
tain techniques were culturally favored at certain times over, the social identity of those involved in distri-
in certain places. John Rowe (1979) has suggested bution networks affected both the value and the
that the term qompi was used by the Inka to refer to meaning of objects produced in the Inka empire
tapestry weave cloth and that tapestry might be re- (Morris 1986; Costin 1993).
garded as a "state" technology Sophie Desrochiers
(1986) has argued that the term refers to all cloth
finished on both sides. Lechtman (1993) suggests Textile Design and Symbolism
that textile technology was central to its meaning and
value in the Inka empire, pointing out that among Late Horizon textiles can be divided stylisti-
the Inka, structural techniques such as tapestry, cally into State Inka, Provincial Inka, and a myriad
double- and triple-cloth, and warp-patterned weaves of hybrid styles. The latter two differ from the
predominated. These are techniques which integrate former in many ways, including their formal (over-
the design into the structure of the fabric such that all dimensions and proportions); technological
structure and design cannot be separated. Ultimately, (method of spinning and plying thread; direction
"proper" Inka garments combined appropriate tech- of warp; type of weave; type of loom; type of
nology with appropriate use of particular designs fiber; thread count); and stylistic (overall design; types
and symbols (see below). of motifs; placement and organization of motifs;
Cloth had a multiplicity of values and mean- amount of standardization; color scheme and pal-
ings in the Inka empire. In a way analogous to ette) characteristics (A. Rowe 1984; 1992). Such di-
Spielmann's (this volume) discussion of how ritually visions have implications for the organization of
powerful objects acquire their power, I identify five production, as they inform us about patterns of
ways material objects acquire their sociopolitical training and access to the knowledge necessary to
meaning and value. These are: create appropriate and meaningful garments.
Studies of surviving examples of tapestry-
1) the rareness and/or inherent value of weave qompi (almost all of them men's tunics \unqu
the materials used to make them; in Quechua]) suggest the design structure of State
2) the amount of labor invested in their Inka-style tunics was based on the use of a concep-
production; tual grid defining square or slightly rectangular blocks
3) the symbolic meanings of the materi- on the tunic field. It has been suggested that the use
als and design motifs; of such a standardized grid facilitated accounting
4) the social identity of the person(s) who for the labor that was invested in textile manufac-
manufactured them; ture (e.g., Durland 1991; Niles 1992), since tribute
5) the social identity of the person(s) who and work obligations in the Inka empire were often
distributed or initially used them. expressed in terms of labor input rather than unit
output (Murra 1975:143-146).
All of these strategies were used to impart value Perhaps the simplest expression of the grid
and meaning to Andean cloth. Many of the materi- was the black and white checkerboard pattern tunic,
als used to make cloth were ubiquitous — especially of which there are at least six published examples
llama wool and cotton — although the state ostensi- (Figure 8.1). Ann Rowe (1978) and John Rowe
bly controlled access to them (Murra 1965). Other (1979) have documented a high degree of standard-
materials used in particularly fine cloth — including ization in the format, metric attributes, manufactur-
alpaca and vicuna wool, metal thread, and tropical ing techniques, and formal design layout among the
bird feathers — were rare and therefore intrinsically extant examples of checkerboard tunics. This sug-
valuable. State control over these materials and gests a relatively high amount of supervision in manu-
sumptuary laws limiting their display reinforced their facture and standardized usage by some categories)
significance. The large amounts of labor that went of persons with relatively large affiliation and simi-
into the production of some of the finest woven lar social status.
pieces -- including the amount of time and number A more complex expression of the grid is the
of hands - clearly increased the worth of this cloth use of design elements called t'oqapu to fill some or
in the value system of the Inka (compare Niles all of the grid blocks. T'oqapu consist of abstract,
1992:51). Beyond their general preciosity, some geometric, and occasionally figurative design mo-
126 Cathy Lynne Costin
design elements may be limited to a band at the waist
(Durland 1991: Figure 15; Reid 1986: Figure 57) or
incorporated into an over-all pattern (Levenson
1991: Figure 451). The extant multiple-t'oqapu tu-
nics have not proven as amenable to analyses of
standardization as have the tunics that use a single
t'oqapu element. While this may be a function of the
comparatively small sample size relative to the num-
ber of t'oqapu elements used by the Inka and to the
apparent diversity of their arrangements, it is more
likely an indication that the muluple-t'oqapu tunics
were woven for specific individuals who possessed
a relatively unique combination of rank and office.
Cloth garments defined people and groups by
indicating aspects of their social identity. Through
regulation and distribution of cloth, the state actively
controlled visual messages and the public percep-
tion of the social identities of subject populations.
The 16th century chronicler Francisco de Jerez
(1917[1534], cited in Morris 1991:522) indicated that
people dressed according to their social status, posi-
tion in the state bureaucratic hierarchy, and the du-
ties they performed. Cloth patterns signaled group
membership and visually distinguished among the
12 royal lineages and between the ruling family and
other nobles (Murra 1962; Morris 1991). Defeated
warriors were stripped of their own garments and
Figure 8.1. Military checkerboard patterned tu- given new ones to indicate their conquered status
nic. Redrawn from Rowe 1979: Figure 1. (Lechtman 1993:256). Elites of newly incorporated
regions were given "gifts" of cloth (Diez de Miguel
tifs. T'oqapu are arranged on tunics in two ways, 1964[1547]; Ortiz de Zuniga 1967 [1562]) as insig-
each with two subvariants. In the first, one or two nia of their new (changed) position within society
t'oqapu motifs are repeated on the garment, either in
a band at the waist (Figure 8.2), or, more rarely, in
an all-over (or nearly all-over) design (Figure 8.3).
Not surprisingly, the extant examples of garments
which use repeats of one or two t'oqapu elements
are highly standardized in terms of design layout
and size (A. Rowe 1978; J. Rowe 1979). A set of
wooden templates recovered from excavations at
the coastal administrative center of Manchan sug-
gest weavers were given instructions on design lay-
out (Mackey 1984:Fig 3). There is an unprovenienced
tapestry tunic fragment in the American Museum
which bears a design layout almost identical to that
found on one of the Manchan templates. Since it is
unlikely that the American Museum tunic came from
Manchan, it is reasonable to assume that tunics of
similar design were manufactured under state su-
pervision in several different parts of the empire.
The homogeneity of single or double-t'oqapu tunics
suggests that, like the checkerboard tunics, relatively
large numbers of these garments were woven gen-
erally for later distribution to a class of persons of
equivalent rank or office.
In the second use of t'oqapu, many design ele-
ments are combined on one garment. As with the Figure 8.2. Diamond t'oqapu waist band tunic.
single-t'oqapu tunics, in muhiple-t'oqapu tunics the Redrawn from Rowe 1979: Figure 10.
Late Prehispanic Andes 127
as ideograms for spinning, twining, and weaving. She
suggests that other foqapu are emblematic of the
ethnic groups from which the weavers were recruited,
drawing an analogy with the makers' marks found
on adobe bricks from the North Coast Moche cul-
ture (Hastings and Moseley 1975). While I disagree
with her specific interpretation that the foqapu rep-
resent textile production techniques and the ethnic
origins of the women who made them, the impli-
cation that the design of these high status garments
reflect the identities and interests of the weavers -
the producers — rather than only those of the secu-
lar and religious elites who wore them is important
for those who analyze the "meaning" of material
things. The significance is this: nonverbal commu-
nication through material culture is not a unitary act
where the artisan is a passive, transparent, nonreactive
conduit channeling the ideas and intents of the spon-
sor of production. Rather, the process that begins
with the desire for or conceptualization of an ob-
ject and ends with its use or display in a series of
acts that involve multiple personae and multiple
voices intermingling in the creation and dissemina-
tion of what we often facilely regard as a singular
"message." Thus, in our search for meaning in ma-
terial culture, we must consider consciously the of-
ten competing strategies and goals of the message
Figure 8.3. "Inka key" t'oqapu tunic. Redrawn creator (here the weavers) and the message bearer
from Durland 1991: Figure 20. (here, the person commissioning or wearing the tu-
nic).
(Morris 1995:431). Thus, social identity was strongly Virtually all of the intact Inka garments that
signaled and manipulated by state controls of dress have survived from prehispanic times are in muse-
to meet state ends. ums or private collections. Few were scientifically
In speculating about what the t'oqapu design excavated or carry reliable provenience. These con-
motifs meant in a literal sense, some archaeologists ditions potentially limit our ability to understand the
and art historians have suggested that the t'oqapu can contexts in which various foqapu designs were used
be "read" as a text of some sort. To this end, they and the exact meaning of variability in terms of motif
have been variously interpreted as a syllabary (Barthel amount, structure, and content. However, we are
1971; de la Jara 1975; Glynn 1981), as heraldic (Rowe not clueless about the social and political significance
1946; Zuidema 1982), as caiendrical (Zuidema 1977, of foqapu structure. One of our primary sources
1982), as indicators of ethnicity and origins (Zuidema of information about the contexts in which vari-
1982), and as pictographic (Reid 1986). A few spe- ously patterned textiles were worn is the early colo-
cific foqapu have recognizable referents. For example, nial writings and drawings of the native Andean
in the famous "royal" tunic at Dumbarton Oaks chronicler Guaman Poma de Ayala (1980[1615]).
(Levenson 1991:Figure 451; Stone-Miller 1997), one The specificity of Poma's drawings and descriptions
foqapu element repeated 33 times accurately depicts suggest he was familiar with garments that belonged
the checkerboard military tunics described above, to particular individuals — including many Inka rul-
complete with the stepped red yoke, neck slit, and ers who "survived" in the form of the royal Inka
zig-zag band at the hem. Stone-Miller (1997), point- mummies which were periodically paraded during
ing to stylistic affinities between some t'oqapu and public ceremonies — and could thus accurately de-
design motifs found on the textiles of earlier Andean scribe their contexts of use and the garments worn
cultures, argues that at least royal foqapu refer to the by particular individuals. John Rowe (1979) has ana-
people and places conquered (and to be conquered) lyzed the contexts in which various design types and
by the empire. configurations are shown in Guaman Poma's draw-
In a thought-provoking analysis, Durland ings. Because Guaman Poma's text often identifies
(1991) has argued that many foqapu designs are ex- the office or social role of individuals portrayed in
plicitly representative of the cloth-making process. the drawings, Rowe has been able to identify con-
For example, she identifies certain design elements sistent associations between certain aspects of tunic
128 Cathy Lynne Costin
decoration and social identity. I summarize his find- and many foqapu on textiles are not found in Guaman
ings here. Rowe suggests checkerboard pattern tu- Poma's drawings. This suggests to me that knowl-
nics are associated with ranking warriors and distin- edge of how to create and specifically "read" these
guished military personnel (Figure 8.4a). A zig-zag symbols was restricted and lost in the first few gen-
line at the bottom-edge of checkerboard tunics dif- erations after conquest, when they lost their relevance
ferentiates between Inka royalty, nobility, and ethni- and utility. Thus, Guaman Poma may have been
cally-Inka persons of rank on the one hand and com- familiar with the general tradition of who wore what,
moners and provincial nobles on the other. This but not with the actual meaning of the symbols, in-
distinction is not just an artistic convention on Poma's dicating the strong dividing line between what the
part; examples of checkerboard tunics with (Rowe general populace knew and what only the royal elite
1979: Figures 2,3) and without (Rowe 1979: Figure and their select artisans knew.
1) an embroidered zig-zag line exist. Many soldiers are Hierarchy and social identity was expressed in
represented in tunics without decorative elaboration, clothing in other ways as well. Low status individu-
suggesting that even the checkerboard may have been als wore undyed clothing. Only higher status indi-
a signal of relative social rank, seniority, and/or ability viduals wore colors, and the number of colors in-
demonstrated in previous campaigns. corporated into a garment also indicated status.
T'oqapu tunics were worn by bureaucrats, Certain colors -- especially red and yellow — were
nobles, and royalty. Low-level bureaucrats wore rela- tied more closely to the state (A. Rowe 1992).
tively simple designs (Figure 8.4b). Multiple t'oqapu Individuals of higher social status with more
tunics ate always associated with persons of rela- roles and greater responsibilities likely owned sev-
tively high social rank or bureaucratic office. The eral tunics with different designs, representing dif-
amount and positioning of t'oqapu seems loosely tied ferent aspects of their social identities or perhaps
to rank as we understand it: in Guaman Poma's changes in social identity over their life span. For
drawings, all-over multiple-/'^/)/* tunics are restricted example, a burial from the Nazca drainage on the
to royalty (Figure 8.4c), although royal figures do south coast yielded two state Inka tunics buried with
not always wear all-over multiple t'oqapu tunics. Un- one individual (Levenson 1991-.Figures 449-450). A
fortunately, many of the t'oqapu that Guaman Poma red and white checked tunic likely represented his
illustrated are not found on extant textiles (indeed, military status, while an Inka key tunic (compare Fig-
some are European symbols, letters, or numbers) ure 8.3) probably represented his role in the state
EIQVARTOCWTAA
Figure 8.4. Guaman Poma's representations of tunic design and social status, (a) Military leader wearing
checkerboard patterned tunic, (b) Low-level bureaucrat wearing t'oqapu waistband tunic, (c) Emperor Topa
Inka wearing all-over t'oqapu tunic. Redrawn from Guaman Poma 1980[1615]: 128, 328, 90.
Late Prehispanie Andes 129
bureaucracy. The fact that the burial also contained cloth at some point in their lives. The early colonial
a mantle woven in a hybrid Inka-Nazca style (A. chroniclers recorded the customary divisions of la-
Rowe 1992:29) suggests this individual retained ties bor along axes of age and gender in the Inka em-
to the local community in which he served/origi- pire. Many of these tasks were associated with tex-
nated. tile manufacture. Young boys and girls herded the
Women did not wear the unqu tunic with sewn flocks of llamas and alpacas that provided much of
seams. Rather, they wore a long rectangular cloth — the fiber. Young girls collected the flowers and herbs
called aqsu — which was wrapped around their body used to make dyes. Adult men and women worked
and held secure by a woven belt at the waist and (at the cotton fields and processed the cotton fiber.
least among elite women) fastened at the shoulders Females of all ages and some age categories of males
with metal pins. Over this dress they often wore a spun thread. All adult women and some adult men
rectangular mantle folded-around the shoulders and wove.
also secured with a (metal) pin. I would like to note In the late prehispanie Andes, cloth produc-
that Guaman Poma illustrated many high-ranking tion figured largely in both the domestic/subsistence
women with what appear to me to be bands of and political economies. I focus here primarily on
foqapu on their belts, at bottom of their aqsu, and on the weaving stage of cloth production, because it
their mantles. Rowe does not comment on the use was at this stage of production of Inka-style cloth
or significance of t'oqapu for women's relative sta- that the greatest labor value was added and mean-
tus.1 ing was encoded. For purposes of this discussion, I
While we cannot yet resolve the explicit, emic identify four broadly defined categories of social
meanings of the state-style t'oqapu, it is clear that tex- persons who made cloth. Commoner women wove
tiles were a powerful form of nonverbal commu- rough cloth ('awasqa), used by their families and given
nication that proclaimed aspects of the social iden- as tribute to clothe and shelter the army. Elite women
tity of those who wore them. As Craig Morris of conquered provinces wove finer cloth (probably
(1991:523) has stated, qompi) that was sent to the emperor in Cuzco each
year. Two named groups of specialists -- one male
It seems probable that the designs of these and one female — produced vast quantities of cloth
tunics were related to the status of the (primarily qompi) directly for state religious, military,
wearers and, in some cases, to the situa- and political institutions.
tions in which they were worn. This would
be consistent with the use of visual insig-
nia of group membership as one of the Part-time "Specialists"
essential functions of garments. The use
of bright and highly visible designs would All women in the Inka empire were cloth-pro-
tell knowledgeable observers at a glance ducing "specialists," since it was their obligation to
the ethnic identity and other essential so- produce thread and cloth beyond their own needs,
cial and political characteristics of the for the use of their households and the state. Women
people they encountered. It would also traditionally made clothing and blankets for their
allow them to quickly assess the compo- immediate families. The Inka state increased their
sition of a large group ... This capacity to work load by requiring that they produce thread and
communicate social information rapidly cloth as part of their tribute obligations. This chusi
was essential, given the variety of the com- and 'awasqa cloth was distributed to the army — sol-
ponents of the empire. diers were entitled to clothing, blankets, and tents —
and to some crews working on state construction
Clothing thus mediated the actions and reac- projects (Betanzos 1987[1551]:61-63).
tions of observers towards the wearers and the in- Commoner "housewives." Women and
teractions among them. As the insignia of rank and cloth production were intimately connected in
office, cloth garments were "keys" to social, eco- Andean societies (cf. Conklin 1990; Moseley 1992;
nomic, and spiritual prerogatives in Late Horizon Gero 1992; Donnan 1995; Costin 1995), and the
society. Consequently, they embodied crucial social Inka administrative structure recognized the impor-
information whose dissemination and ownership tance of all women in meeting the extremely high
needed to be controlled and limited. demands for cloth to be redistributed by the state.
In his delineation of female census categories, the
early colonial chronicler Guaman Poma listed mar-
CLOTH PRODUCERS ried women first.2 Women of this category were
called awakuq rvarmi in Quechua. Guaman Poma
Virtually everyone in the Inka empire had a (1980[1615]:190) glossed the term in Spanish as
role — albeit often minor ~ in the production of texedora [sic], or weaver. Guaman Poma
130 Cathy Lynne Costin
(1980(1615]: 190) stated that the primary responsi- ductive and reproductive lives - at least from the
bility of women in this census category was to spin point of view of the state — was as makers-of-cloth.
and weave "para el Ynga y demas senores capac Guaman Poma's discussion of the awakuq
apoconas y capkanes y para soldados" (for the Inka warmi is not the only indication in his work of the
and also for powerful men, captains, and soldiers importance of cloth production to female identity.
[translation by the author]). It is apparent from the Of the 10 female census categories enumerated by
rest of text that the category included all women of Guaman Poma, eight (ranging in age from 5 to more
this age — roughly 25-50 years old — and not some than 80 years old) emphasize textile production-re-
limited specialist group of weavers. These women lated tasks among their primary economic responsi-
were clearly identified as tribute paying women bilities (Table 8.1). The text discusses in detail the
("mujeres de tribute"). Taking apart the term, we type(s) of thread and/or cloth each category of fe-
see that warmi translates as married woman, and males was expected to produce and for whom the
awakuq may be derived from the same root as goods were produced (i.e., family, "community,"
'awasqa, or rough cloth. The primary identity of this state institutions).
large category of women in the prime of their pro-
GUAMAN POMA'S
SPANISH GLOSS
CENSUS CATEGORY (ENGLISH) AGE CLOTH PRODUCTION TASKS
VI: Qhuru Thaski de cabello corto 12-18 learn to spin and weave
(short-haired) herd camelids
VII: Pawaw Pallaq que recoge flores 9-12 collect flowers for dye
(flower collectors)
Table 8.1. Inka female census categories and tasks related to textile-production. After Guaman Poma
1980[1615]: 190-209.
Late Prehispanic Andes _ _ 131
One of the most valuable components of illustrated with images of girls or women engaged
Guaman Poma's work is the more than 450 line in activities related to cloth production: two weav-
drawings he made to illustrate his text. The accu- ing, two spinning, one spinning while herding lla-
racy of these drawings has been noted by modern mas, and one collecting plants for dyes (Figure 8.5).
scholars (e.g., Adorno 1986). The image Guaman Of the drawings that do not make reference to tex-
Poma used to illustrate his discussion of the arvakuq tile-related activities, one illustrates the category of
tvarmi is a woman weaving (Figure 8.5a). The em- women over 80 years old and the other three repre-
phasis on the female role in textile production and sent girls under nine years of age. Thus, only the
the symbolic association between females and cloth very youngest and very oldest females were not rep-
is reinforced in other drawings by Guaman Poma resented in a way that clearly associated them with
that accompany his text on female roles and respon- cloth production.
sibilities. Six of the ten female census categories were
Figure 8.5. Guaman Poma's representations of feminine labor and cloth production, (a) awakuq warmi
weaving; (b) payakuna weaving; (c) unquq k'umu spinning; (d) sipaskuna spinning; (e) qburu tbaski herding
camelids; (fypawawpaJ/aq collecting flowers for dye. Redrawn from GuamanPoma 1980(1615]: 191,193,196,
198,200,202.
132 Cathy Lynne Costin
In contrast with the characterization of female activities associated with cloth production. The lack
labor in Guaman Poma's treatise on Inka society, of association between males and textile produc-
only four of the ten male census categories listed tion is underscored in the line drawings that accom-
tasks related to fiber procurement, processing, or pany the text. Only one of the ten male census cat-
weaving as part of their basic roles and responsibili- egory illustrations depicts an activity related to cloth
ties (Table 8.2). Of these four categories, three listed production. He is a 12 year old boy shown trap-
the activity of herding camelids. This contribution ping birds while herding. Discussions in other colo-
was not related exclusively to cloth production, as nial documents confirm the limited male role in non-
these animals were quite important in the Andean specialist textile production (e.g., Diez de Miguel
world as pack animals and as a source of food. In 1964[1547]:61).
addition to herding, prepubescent boys collected There is no question but that the commoner
feathers and some were expected to spin. Ill and women of the Inka empire — especially the awakuq
handicapped men were obliged to weave. Impor- warmi — produced enormous quantities of doth for
tantly, men in the prime of their productive and re- the state while performing their tribute-labor obli-
productive lives were not required to participate in gations. Using the most conservative of estimates,
GUAMAN POMA'S
SPANISH GLOSS
CATEGORY (ENGLISH) AGE CLOTH PRODUCTION TASKS
Table 8.2. Inka male census categories and tasks related to textile-production. After Guaman Poma 1980[1615]:
170-189.
Late Prehispanic Andes 133
it seems likely that hundreds of thousands of pieces used by those weavers. At the time of Spanish con-
of 'awasqa could have been mobilized through this quest, state warehouses were filled with spun thread
form of dispersed corvee each year. Assuming a (as well as raw fiber and finished cloth), suggesting
population of 8-10 million people in the Andes at thread was mobilized from the tributary popula-
the time of the Spanish conquest, an estimate of 1 tion, presumably women. Lack of reference to this
million households is highly conservative. The work likely made it largely invisible in the scheme of
chroniclers report that each household was required social recognition and importance.
to produce one blanket and one tunic each year, al- Elite Provincial Women. In the highland
though yearly tribute requirements likely varied geo- chiefdoms of pre-Inka times, elite household-based
graphically and over time (LeVine 1987; Costin 1993). (female) labor may have been the primary source
Given that the size of the Inka army stood in the of the elaborate, labor intensive cloth that circulated
hundreds of thousands and the large number of in local and regional political economies. The
men that were to be remunerated with cloth after ethnohistoric documents describe women weaving
serving on state construction and mining projects, it in polygynous elite households (e.g. Polo de
probably did take all the households in the empire Ondegardo 1940[1571]:141; Falcon
to provision the state with rough cloth. Women 1918[1565]:154), and my analyses of archaeological
pursued this work in a domestic context using their data from the Upper Mantaro Valley indicate weav-
own local, traditional techniques (Costin et al. 1989). ing was much more intensive in elite households than
No additional skills beyond those expected of all in commoner households before the region was
women were required to produce this cloth. The conquered by the Inka empire (Costin 1993). After
undecorated 'awasqa cloth collected as tribute car- conquest, the intensity of weaving in elite households
ried little symbolic load — beyond the significance did not increase as much as it had in commoner
that it was redistributed through the state redistribu- households. In commoner households, debris in-
tive channels and represented expropriated female dicative of textile production increased more that
labor — and was little standardized. Although dis- 300% after Inka conquest; in elite households it in-
tributed by the state, it was essentially the same stuff creased by slightly less than 50%.
people had at home. The greatest impact of Inka tax demands on
I have dealt elsewhere with the effects of Inka elite women was the change in distribution networks
conquest on the productive labor -- as it was ex- for the cloth they wove and in the new symbolic
tracted in thread and cloth — of commoner women and legal relations among producers and consum-
(Costin 1993, 1995, 1996b). Notably, my archaeo- ers. I have argued elsewhere that after the Inka con-
logical data from the Upper Mantaro Valley suggest quest the products of elite women's labor were at
they worked significantly more on textile produc- least partially redirected from household and lineage
tion-related tasks than they had before they were use to state ends (Costin 1996b). Elite men who
forcibly incorporated into the Inka empire. Tax- were recruited into the Inka bureaucracy were en-
related demands on their labor were not compen- titled to cloth from the state; the ethnohistoric records
sated by reductions in other obligations or expecta- suggest that some of this cloth was made by their
tions. Commoner women appear to have benefit- wives and they were "allowed" to keep it (Mackey
ted little in material terms from the enormous contri- 1984). Legally, however, no one was allowed to
bution they made to the political economy. There is wear qompi unless it had been granted by the em-
no evidence that any prestige accrued to them for peror. Thus, although this cloth was transferred
their "contributions" to the state. This stands in con- physically as it had been before the conquest (i.e.,
trast with many commoner men, who received cloth within the household), under Inka domination the
during their military service (Murra 1962; Spurling transmittal of this cloth from weaver to wearer was
1992) and extra rations of highly valued maize or mediated symbolically and legally by the state.
chicha (maize beer) when they performed corvee la- The ethnohistoric documents also describe
bor for the state (Hastorf 1991). "presents" of doth made by the wives and daugh-
It is likely that much of the female labor con- ters of provincial elites which were give to the em-
tributed to state textile production went socially un- peror and his representatives (Polo de Ondegardo
recognized. Despite a strong emphasis on the Inka 1940 [1571]:146; Rowe 1979). These gifts are treated
cloth industry in the colonial descriptions, there is as though they were nearly obligatory, striking me
little discussion of spinning. There are no terms for that they were in fact part of the tribute obligations
"spinning" or "spinners" in early Quechua dictio- of the conquered populations, although they were
naries (Spurling 1992:244) and there is no term for not technically labeled as such. It is somewhat un-
spinning specialist among the detailed lists of spe- clear exactly what kind(s) of cloth elite provincial
cialists compiled by Falcon (1977 [1571]) and Munia women wove for "gifts" to the state. The Inka em-
(1946[1590]). Discussions of state weaving special- perors impressed conquered populations by appear-
ists (see below) rarely identify the source of thread ing in local dress when they visited the provinces
134 Cathy Lynne Costin
(Saramiento 1960[1572]:249; Betanzos rewards to some men in return for exemplary mili-
1987[1551]:181; Murra 1962:721) suggesting that tary or bureaucratic service to the state; and as pro-
some of the cloth intended for the emperor as gifts ducers of goods important in political and religious
or tribute was woven in indigenous pre-Inka styles. rituals, especially maize beer and cloth.
Rowe (1979:239-240) suggests that provincial elite The aqllakuna are everywhere asserted to be
women may have consciously created distinctive gar- female. They were recruited into state service at a
ments in order to catch the attention and favor of relatively young age, about 10 years old. Those who
the emperor. These relatively unsupervised, partially were not chosen as sacrifices or given as rewards
"Inkaniaed" women may also account for the sur- ostensibly lived out their lives as aqllakuna in special
viving textiles that have been classified as stylisticallyfacilities called aqllawasi ("house of the aqlla"). The
Provincial Inka or Inka-related hybrids of local and term wasi itself may tell us something about aspects
state styles. Although the chroniclers mention the of the social identities of the aqllakuna, as in addi-
"favor" bestowed upon the wives and daughters tion to "house" it can also mean storehouse (espe-
of elites who sent these "gifts," the change in the cially for valuables such as chicha) or jail (Guaman
nature of the relationship between producer and user Poma 1980[1615]:1105-1106). The chronicler Munia
from a kin-based to a class (state) based system un- (1946[1590]: 191) specifically referred to the aqllawasi
doubtedly affected the status of and rewards en- as storehouses. Several compounds identified as
joyed by provincial elite women. aqllawasi have been investigated archaeologically at
Inka provincial centers (Morris 1967, 1974; LeVine
1985; Morris and Thompson 1985). The com-
Full-time Weaving Specialists pounds are highly distinctive, with small, cell-like
rooms, limited access to the rest of the community,
In addition to the vast quantities of cloth ex- and high densities of artifacts related to textile pro-
tracted as tribute from the women of conquered duction.
households, the Inka procured large amounts of The aqllaivasi seems to have been a place where
cloth through patronage of two groups of special- female sexuality stood in marked contrast with that
ist weavers whose products went exclusively to the of married commoner women. Those aqllakuna
state. These two groups represent a set of stark who were not sacrificed or given out as wives or
contrasts in terms of social identity (Table 8.3). concubines — in other words, those who remained
Aqllakuna. Aqlla (plural aqllakuna) means state weavers — never married; they were expected
"chosen female." The aqllakuna who wove for the to remain virgins for life. Loss of virginity by a
state were a subset of a group of women recruited sequestered aqlla was punishable by death for both
to serve three key services to the state: as sacrificial the woman and her lover (Cieza 1959 [1553]:21-22;
victims; as women specially trained in the feminine Moore 1957:165). Female "victims" of acts defined
arts (weaving and brewing) who were distributed as by Inka state law as sexual crimes, such as rape, in-
AQLLAKUNA QOMPIKAMAYOC