Cotton Spinning Technology - Mexico

The Development of Prehispanic Cotton-Spinning Technology in
Western Morelos, Mexico
Michael E. Smith
Loyola University of Chicago Chicago, Illinois
Kenneth G. Hirth
University of Kentucky Lexington, Kentucky
Cotton textiles played important economic, social, and political roles in the Prehispanic civi­lizations of Mesoamerica, yet archaeologists have made little progress in the analysis of tex­tile production beyond the simple identification of spindle whorls. In this article we identify and describe whorls and a second artifactual marker of cotton spinning: ceramic spinning bowls. Quantitative changes in the occurrence of these artifacts in excavated contexts from the Mexican state of Morelos are then used to discuss the development of the local cotton in­dustry from the Epiclassic through the Late Postclassic periods. The Postdassic cultures of Morelos apparently produced a number of innovations in cotton-spinning technology, and these changes along with increased textile production levels are linked to demographic, eco­nomic, and political changes occurring throughout Central Mexico at that time.
Introduction
Archaeologists and ethnohistorians studying Prehis­panic cultures in Central Mexico (FIG. 1) have long rec­ognized the importance of cotton as a major resource and tribute item among contact period groups (e.g., Anawalt 1981; Berdan 1987). Although archaeologists have de­scribed ceramic spindle whorls used in the production of cotton cloth (e.g., Parsons 1972, n.d.; Wiltfang 1975; Meighan 1976: 77-81), they have been slow to identify additional artifactual evidence of textile manufacture or to examine technological or quantitative changes within spe­cific Prehispanic textile industries. This report discusses material evidence for Prehispanic cotton spinning in west­ern Morelos, an important cotton-producing area of the central Mexican highlands. Diachronic changes in spin­ning technology and intensity are documented and inter­preted, leading to suggestions concerning the changing role of cotton textiles in the Postclassic cultures of Central Mexico.
Mesoamerican Textile Technology
Cotton thread was spun by hand in Prehispanic Me­soamerica. Women worked either alone or in groups pro­ducing thread that was woven into cloth on the traditional backstrap loom (Anawalt 1981). Three simple tools were required to produce cotton thread: 1) a thin wooden spindle fitted with, 2) a ceramic disk or whorl, and 3) a small ceramic bowl used to support the spindle as it twirled. These three items are mentioned and illustrated (FIG. 2A) in two lists of textile equipment by Sahagin (1950-1969, Book 6: 201, fig. 30, and Book 8: 49, fig. 75). In Sahagun's terminology, the spindle and whorl are referred to together as malacatl, while the bowl is called tzavalcaxitl.' Other conquest period illustrations of these items are found in Duran (1967, I: fig. 28) and the Codex Mendoza (1980: Folios 58r, 59r, 68r); the latter example is reproduced here in Figure 2b.
Ethnographic descriptions and illustrations of spindles, whorls, and bowls used for spinning are abundant in the literature on modern Mesoamerican textiles (e.g., Cordry and Cordry 1968; O'Neale 1945: 8-10). Whorls used in
1. The Nahuatl term malacatl derives from malina, to twist, and acatl, cane or reed. Separate terms for spindle and whorl are not found in conquest-period sources. Among contemporary Nahuatl-speaking weav­ers in the Sierra Norte de Puebla, Mexico, malacatl refers to both the spindle and whorl, although separate terms do exist: malacatitl for the whorl and malacapuiotl for the shaft (Frances Berdan, personal com­munication). The 16th-century term tzemalcaxitl derives from tzahua, meaning to spin, and caxitl, a pottery bowl. This term has not been reported among modern weavers, probably because there is no special vessel used specifically for spinning. Related terms for specialized spin­ning tools mentioned in Sahagin are teucuitlazaoalcaxitl, meaning golden spinning bowl (Sahapin 1950-1969, Book 9: 18) and xaxalmalacatlic iujtanatl, which is a shallow spindle used to spin with feathers (Sahagün 1950-1969, Book 8: 49).


spinning cotton are usually ceramic, although wood, stone, and gourd examples are also reported. Whereas Aztec spinning bowls were a functionally-specialized vessel type, modern cotton spinners tend to use broken ceramic or gourd bowls to hold the twirling spindle. These bowls are needed during the thread making process to support the spindle when spinning short-staple fibers like cotton. Medium and long-staple fibers like wool or maguey are best spun with a drop-spinning technique, where the spin­dle is dropped by hand and allowed to spin freely, sus­pended only by the twisting thread. While it is possible to spin cotton in this manner, a far higher degree of control is obtained when the spindle is supported (Hoch­berg 1980: 19-44). The use of a small bowl for support of the spindle makes the process easier and more efficient (Cordry and Cordry 1968: 27; O'Neale 1945: 10). It is interesting to note that while spinning bowls are found in nearly all Prehispanic illustrations of spinning in Central Mexico, examples in the Mixtec codices from highland Oaxaca show women spinning without the use of a sup­port bowl (e.g., Codex Nuttall 1975: 11).
The Morelos Data
Archaeological Evidence for Spinning
Spindle whorls and spinning bowls were made of ma­terials that can be recovered through archaeological inves­ tigation. Spindle whorls occur as formed ceramic disks whose function can be inferred by direct analogy from ethnohistoric and ethnographic usage. In a study of Aztec-period whorls from the Teotihuacan Valley, Mary Parsons (1972) notes a bimodal distribution of whorl size and weight. She interprets the smaller whorls (her type III) as having been used for spinning cotton, and the larger whorls (types I and II) as having been employed in spin­ning maguey fiber. Parsons' functional interpretation of whorl size is based on ethnographic analogy (O'Neale 1945; Kent 1957) and is technologically reasonable. Whorl weight is a major determinant of the angular mo­mentum (the speed and force of spin) of the twirling spindle, and short fiber materials such as cotton require less angular momentum than does the longer fiber ma­guey.
This bimodal distribution of whorl sizes is also found among archaeological spindle whorls from the state of Morelos. Weights of 85 unbroken whorls from the site of Teopanzolco in Cuernavaca dating to the period A.C. 1150-1400 are plotted in Figure 3. These whorls range from 2 g to 74 g in weight, with most examples clustering between 2 g and 18 g. Following Parsons (1972: fig. 13), we interpret whorls under 18 g to have been employed in spinning cotton and whorls over 34 g to have been used in processing maguey; whorls with intermediate weights (18-34 g) may also have been used in spinning maguey although this assessment remains tentative for the time being. It is important to note that there are significant stylistic differences in whorl decoration that correlate with their separation into functional groups of cotton- and maguey-spinning whorls. Within the Teopanzolco collec­tion nearly all of the whorls under 18 g in weight exhibit plain smooth surfaces while all but one of those over 18 g in size are decorated with incised or mold-impressed designs. The Teopanzolco collections are representative in terms of the forms and sizes of Postclassic whorls found throughout Morelos.
Nearly all of the small cotton-spinning whorls known from Morelos conform to Parsons' type Ma plain variety (1972: 52-53). Several examples from Teopanzolco are illustrated in Figure 2c. Maguey whorls are less common in Morelos than in the Basin of Mexico, probably because the species of maguey employed for fiber in the Basin (the so-called pulque magueys, Agave mapisaga andA. salmiana [Gentry 1982: 594-608]) do not grow in most of the state because of its lower elevation and warmer climate (there are relatively more large whorls at Teopanzolco than in areas of lower elevation to the south of Cuernavaca). Similarly, the majority of the large whorls recovered in the state have mold-impressed geometric designs, resem-


bling examples from the Toluca Valley illustrated by Par­sons (1972: plates 15-16).
Prehispanic spinning bowls correspond to a group of small, orange ceramic bowls with supports, which have been recovered in Late Postclassic contexts in the Basin of Mexico and surrounding areas of Central Mexico (FIG. 2E, F). Typological studies based on paste and surface dec­oration (Parsons 1966; Smith 1983) place these bowls within the well-documented ceramic category known as Aztec III Black-on-Orange which dates to the Late Aztec phase of A.C. 1350-1521+ (TABLE 1; Parsons 1966). The first published reference to Aztec spinning bowls appears to be Bernal's mention of "Unas copitas de las llamadas `bases de malacate'" (1948-1949: 36), although Bernal questions this interpretation and calls the vessels toys. Medellin Zenil (1952: 56, 1960: 200), in his discussions of Aztec pottery recovered in Veracruz, accepts without comment the textile-related function of these bowls, but subsequent discussions of Aztec pottery and textile pro­duction have not followed up this important lead. Med­ellin Zenil's functional interpretation is apparently based on his association of Aztec III miniature bowls with those portrayed in conquest period pictorial codices (FIG. 2A, B).
Confirmation for the use of these miniature bowls as spinning vessels comes from three properties of the bowls themselves. 1) Aztec III miniature bowls have a relatively narrow range of orifice diameters, suggesting their man­ufacture for a specific use with technological constraints² (FIG. 4). 2) They always have supports, usually annular (FIG. 2E), with some examples exhibiting slab supports (FIG. 2F), that provide stability for their use in spinning. 3) Many examples of these bowls examined by Smith exhibit patterns of abrasive wear on their interior base, which could have been produced by their prolonged use in supporting the twirling spindle. Furthermore, the Aztec III miniature vessels are orange, the color used to depict spinning bowls in the ethnohistorical sources (e.g., Codex Mendoza 1980: Folios 58r, 59r, 68r); they are sometimes referred to as "golden spinning bowls" by Sahaglin (see note 1).
While Aztec spinning bowls have been recovered from Late Postclassic deposits around Mexico City (S6journ6 1983: figs. 92, 113-114; Parsons 1966: figs. 39-40), they are not very common in the Basin of Mexico. By contrast, these vessels are quite common in surface collections and
2. Although the frequencies are low, the data in Figure 4 may suggest a bimodal size-distribution. Apart from size, the three tiny bowls are no different from the larger spinning bowls. They could indicate spinning of particularly fine thread that required greater control than the standard cotton thread, such as the Mexican brown cotton with its very short fiber length. excavated contexts in Morelos. Nine whole vessels were recovered from Late Postclassic (Cuauhnahuac phase) burials at the site of Coatetelco (Raul Arana, personal communication), and fragments of these vessels were also represented in Hirth's surface collections from 22 of 71 Late Postclassic residential communities in the Coatlan­Tetecala region in western Morelos. Their occurrence in the nearby Toluca Valley is documented by 19 complete examples contained in the Frederick Starr collection of several hundred ceramic vessels from Postclassic burials in that area (the collection is housed in the Field Museum of Natural History in Chicago). Additional examples are reported from the Aztec garrison of Quauhtochco (mod­ern Huatusco) and from Quiahuiztlan in Veracruz (Med-
Figure 4. Rim diameters of local and Aztec spinning bowls. Local bowls are from excavated contexts at Xochicalco and Coatetelco (Tern­azcalli and Cuauhnahuac phases); Aztec bowls are from Coatetelco and the Frederick Starr collection of vessels from the Toluca Valley at the Field Museum of Natural History.

ellin Zenil 1952: 56, 1960: 200) as well as from Coixt­lahuaca, Oaxaca (Bernal 1948-1949: 39). Although quantitative assessment is not possible, it appears that Aztec spinning bowls are less common within the Basin of Mexico than in those areas outside the Basin that were paying heavy tribute in cotton textiles to the Triple Alli­ance at the time of the Spanish conquest (Codex Mendoza 1980: Folios 23r, 23v, 35r, 48r).
Other types of miniature tripod bowls recovered ar­chaeologically in western Morelos probably also func­tioned as spinning vessels. Small hemispherical bowls with coarse-sand-tempered, local orange paste and a black-smudged surface finish (FIG. 2G) are the same size as the miniature Aztec III bowls (FIG. 4, top). They nearly always have solid conical supports, although slab supports (but not annular bases) also occur (see Smith 1983: 439-441). Like the Aztec III bowls, many of these black-smudged miniatures exhibit wear on their interior bases. Although the latter have not been reported outside of Morelos, Moedano Koer (1941: 30) suggests that other miniature tripod bowls served the same function at Tzintzuntzdn (Moedano Koer 1941: fig. 4). Likewise, Lister (1955: 21, 88-89) discusses the distribution of miniature tripod ves­sels in western Mexico and suggests that their occurrence outside Tzintzuntzdn is associated with the Tarascan ex­pansion.³
Changes Through Time
Archaeological investigations have established that cot­ton was grown throughout Central Mexico by 3400 B.C., if not earlier (Stephens 1967: 260). The earliest spun-cotton yarn was recovered from Purron cave in the Te­huacan Valley and dates to between 1200 and 1500 B.C. (MacNeish, Nelken-Turner, and Johnson 1967: table 30; MacNeish and Garcia Cook 1972: fig. 41). The earliest spinning paraphernalia may have consisted of perforated ceramic disks, made from broken and reworked ceramic vessels, that could have been used as simple spindle whorls.⁴ Reworked ceramic disks dating from 825 to 1025
3. Recent excavations by Smith at Late Postclassic sites in Morelos yielded many examples of local and Aztec spinning bowls and spindle whorls that will allow more detailed descriptions and analyses of these artifacts based upon larger samples than are currently available.
4. While some archaeologists assert that perforated circular disks man­ufactured from potsherds could not serve as spindle whorls (Caso, Ber­nal, and Acosta 1967: 436), ethnographic examples of this practice are known (Weigand 1969: 24). Kent (1957: 473) examined this issue and concluded that while rough asymmetrical sherd disks will not work effectively as spindle whorls, well-made symmetrical examples can serve
B.C. have been recovered in the Tehuacin Valley (MacNeish and Peterson 1972: 187) and continued in use up until ca. A.G. 700, when specially-manufactured whorls appeared for the first time in prehistoric sites throughout Central Mexico.
Excavations of the Xochicalco Mapping Project in 1977 (Hirth 1983) recovered a number of specially-manufac­ured spindle whorls from Epiclassic (A.c. 700-900) con­texts at Xochicalco and Tlacuacingo (FIG. 2D). These ar­tifacts, together with contemporary examples recovered from Cerro Tenayo (Rattray 1966: 130-131) and Cholula (Muller 1978: fig. 24), represent the earliest securely-dated spindle whorls in the central Mexican highlands. Classic-period whorls are found in the Maya lowlands (Willey 1972: 84-86; Sheets 1978: 61-63) and in west­ern Mexico (Meighan 1976: 77-81), but these are spindle whorls stylistically distinct from those of the central high­lands. Formed whorls are absent from Classic deposits at Teotihuacan, although perforated sherd disks are present (George Cowgill, personal communication). Perforated ceramic disks are also present at Xochicalco and Tlacu­acingo (FIG. 211). On the basis of size and weight we believe that both the Epiclassic whorls and disks recovered from western Morelos were used to spin cotton.
The Early Postclassic period in Morelos corresponds to the time period A.C. 900-1200. The early part of this period is poorly understood, and no spinning artifacts have been recovered. The later part of this period corre­sponds to the Tilancingo phase (A.c. 1100-1200) when there was a clear separation between small undecorated cotton whorls and large mold-impressed maguey whorls that continued through the Spanish conquest (see above). Excavated examples of these whorls come from dated de­posits whose ceramics were not fully quantified, so that no judgments can be made regarding their relative fre­quency (TABLE 2). Examples also have been recovered from surface collections in the Coatlan-Tetecala region, where both maguey and cotton whorls have been recov­ered from single-component domestic sites dating to the Tilancingo phase.
Spindle whorls are numerous elsewhere during this pe­riod, including Tula (Franco 1956; Parsons n.d.) and Cholula (Muller 1978: figs. 24, 25). With regard to the whorls at Tula, there are strong stylistic similarities to contemporary whorls from the Huaxtec region on the
as whorls (see Cordry and Cordry 1968: 27; Willey 1972: 80-81). Whether reworked, perforated ceramic disks were actually used as spindle whorls and/or had subsidiary functions will need to be established with future research. While we believe that some ceramic disks were used as whorls we do not believe that this was the only function these disks performed.
Gulf coast (Eldiolm 1944: 459). Both large (maguey) and small (cotton) whorls are found at Tula (Parsons n.d.) and Cholula (Muller 1978: fig. 25) during this period, whereas nearly all of the Huaxtec examples are of the small variety. We suspect that while a Huaxtec-derived whorl tradition was emerging in the area around Tula, a separate and independent highland tradition of small, plain whorls was emerging in Morelos.
Miniature, black-smudged spinning bowls make their first appearance at Xochicalco and other sites in western Morelos during the Temazcalli phase (A.c. 1200-1350). Maguey and cotton whorls are also present, and the cot­ton-spinning artifacts (small whorls and bowls) comprise 0.17% of the ceramic assemblage from this phase (TA­BLE 2). Although there are only five single-component sites in the Coatlan-Tetecala region during the Temazcalli phase, at least one of these produced a cotton whorl from surface collections. In the Basin of Mexico, O'Neill (1962: 269, 277) reports six cotton and two maguey whorls in Early Aztec levels at Chalco; the cotton whorls comprise 0.12% of the ceramic materials for that phase.' Early Aztec surface collections from Huexotla and Xico contain both cotton and maguey whorls in small quantities (Brumfiel 1986). The Chalco whorls appear to include at least one Huaxtec (or Toltec?) style example; other whorls corre­spond to various of Parsons' types, although type IIIa, which she refers to as the Morelos-type cotton whorls, is not present (O'Neill 1962: 217).
Aztec III Black-on-Orange miniature spinning bowls make their first appearance in western Morelos at the beginning of the Cuauhnahuac phase (A.c. 1350— 1521 + ). Cotton-spinning artifacts increase in frequency from 0.17% to 0.59% of the ceramic assemblage recov­ered from Xochicalco and Coatetelco (TABLE 2). Whorls or bowls were recovered at 60% (26 of 43) of the single-component Cuauhnahuac-phase sites in the Coatlan-Te­tecala survey region. A similar increase in cotton whorls is reported for Chalco, where 13 cotton whorls were re­covered, comprising 0.41% of O'Neill's Late Aztec ce­ramic sample (1962: 269, 277). At Huexotla and Xico, however, Brumfiel (1986) reports a 50% decline in both cotton and maguey whorls between the Early and Late Aztec periods. Within the Chalco collections all the small,
5. The Chalco data are taken from O'Neill's (1962) dissertation. Lev­els 1-8 of his test pit have significant amounts of Aztec III Black-on­Orange ceramics with small numbers of Aztec I and Aztec II sherds; these are interpreted as a Late Aztec deposit with some mixture of Early Aztec materials. Levels 9-28, with Aztec I and Graphite Black-on-Red (but no Aztec III) are interpreted as an Early Aztec deposit. Sherd totals for the calculation of the percentages are taken from O'Neill (1962: ta­ble 2).

undecorated cotton whorls and spinning bowls date to the Late Aztec period (O'Neill 1962: 214).
Implications
While cotton textiles were clearly manufactured and used in Central Mexico from a very early date (Smith and Kerr 1968), the appearance of specially-formed ceramic spindle whorls during the Epiclassic period may relate to an increase in the scale and intensity of cotton production. At the time of the Spanish conquest the use of cotton apparel was a privilege reserved for members of the elite, and cotton textiles had become important tribute items within expanding conquest states (Barlow 1949; Anawalt 1981; Berdan 1987). Furthermore cotton was used in the manufacure of the ichcahuipilli (defensive armor) worn by warriors (Penafiel 1985: 17).
The Epiclassic period was characterized by two impor­tant sociopolitical trends—the emergence of numerous small, independent city-states during the breakup of the Teotihuacan empire after A.C. 650, and an increase in the general level of conflict and militarism as a result of in­creased competition over scarce resources within Central Mexico. The demand for Morelos cotton would have in­creased with the fragmentation of Teotihuacdn's interre­gional trade networks. Growth in the number of inter­mediate-size administrative centers during this period (Sanders, Parsons, and Santley 1979) would have raised the demand for cotton textiles proportional to the increase in the number of rural elite. At the same time, an increase in regional conflict would have raised the demand for cotton needed to produce ichcahuipilli. The result would have been an increase in the aggregate demand for cotton at the same time that non-local cotton produced in areas such as the Gulf Coast may have been less readily available due to decreased interregional trade. The appearance of more specialized spinning equipment within Epiclassic sites in Morelos may reflect an increase in the demand and scale of cotton production in Central Mexico.
The data presented above suggest that the peoples of western Morelos produced a number of technological in­novations in cotton-spinning technology during the sub­sequent Postclassic epoch. Local spinning bowls are pres­ent in Morelos around A.C. 1200, while the Aztec and Tarascan variants do not appear until after A.C. 1350. The overall similarity of Morelos and Aztec examples suggests that the latter were patterned after the former. Similarly, the small, plain cotton whorls appear to have their origin in Morelos, where they go back to the Epiclassic period. Plain cotton whorls have not been found in the Early Postclassic deposits at Tula (Parsons n.d.), and at Chalco they do not appear until the Late Aztec phase (O'Neill 1962: 213-218).
The Middle to Late Postclassic transition witnessed a dramatic increase in the intensity of cotton-thread pro­duction in Morelos, as measured by the frequency of spin­ning artifacts in excavated deposits (an increase of 350%) and the prevalence of such artifacts in surface collections from single-component sites. At recently-excavated Late Postclassic sites near Xochicalco, spinning artifacts are both abundant and ubiquitous. Preliminary indications suggest that cotton spinning was carried out by all house­holds, both elite and commoner. The quantities of finished textiles would have been quite high, and in addition to their local use many textiles must have been exported from this area (see note 3). Ethnohistorical documents reveal a heavy trade in Morelos textiles (e.g., Duran 1967, II: 23), and these items are prominent in lists of tribute paid by Morelos polities to the Aztec empire (Codex Mendoza 1980: 23r, 24v).
This active trade in cotton textiles was not limited to Morelos. Berdan (1987) and Anawalt (1981) review eth­nohistorical documentation of similarly high levels of tex-
tile production and exchange throughout Late Postclassic Central Mexico where cotton textiles had four primary uses. First, cotton garments were used as clothing. Al­though the chroniclers suggest that use of cotton clothing was restricted to the elite (see Anawalt 1981), this may have applied only to the Basin of Mexico where cotton was an import. A second utilitarian function was in the manufacture of ichcahuipilli or armor (Periafiel 1985: 17; Berdan 1987). A third, symbolic and sociopolitical, use of cotton is represented by the role of elaborate, decorated textiles in elite exchange, politics, and status validation (see Anawalt 1981). Fourth, cotton textiles served as a medium of exchange in both the tribute and market sys­tems (Berdan 1987). Drennan (1984) has stressed the importance of textiles in transferring economic value over large distances in Prehispanic Mesoamerica because of their high labor investment and low weight. In sum, cot­ton textiles played a number of important roles in Late Postclassic society and economy, and as a result they were manufactured and exchanged at very high levels.
The Middle to Late Postclassic increase in cotton pro­duction in Morelos indicated by the above data contrasts with, but may be related to, the decline in spinning activity reported at Huexotla and Xico in the Basin of Mexico (Brumfiel 1986). Recently, scholars have argued that pro­cesses of population growth, urbanization, and political centralization led to increasing emphasis on agricultural production within the Basin, with a corresponding dis­placement of non-agricultural production to exterior pro­vincial areas like Morelos (Brumfiel 1980, 1986; Blanton and Feinman 1984; Hassig 1985). This model, developed primarily from ethnohistorical documentation of condi­tions under the Aztec empire in the years A.C. 1430-1519, suggests that the observed changes in the regional config­uration of craft production were most dramatic between the pre-imperial and imperial periods, and the authors cited above link the economic changes to the growth and expansion of the Aztec empire.
Our data suggest an alternative interpretation. If the consensus dating of the Middle to Late Postclassic tran­sition at ca. A.C. 1350 is valid (Sanders, Parsons, and Santley 1979: 466-471; Smith 1987), then the shift in the locus of textile production at that time took place nearly a century before the formation of the Aztec empire. Our data then suggest little change in the intensity of textile production between the Early and Late Cuauhna­huac phases (TABLE 2), which correspond to the intervals before and after Morelos was incorporated into the Aztec empire (Smith 1987). Thus the Middle to Late Postclassic changes in regional patterns of textile production cannot be due to the effects of the Aztec empire; however, the three factors singled out to explain the economic changes—population growth, urbanization, and political centralization—are still likely candidates. These develop­ments were not limited to the ethnohistorically-docu­mented imperial period; settlement pattern data show that they had a longer trajectory during the Postclassic epoch (Sanders, Parsons, and Santley 1979). Rather than view­ing the changes in craft production as responses to the development of the Aztec empire, we see both the eco­nomic shift and the rise of the empire being generated by long-term demographic, economic, and social changes in Postclassic Central Mexico. Advances in our understand­ing of the techniques of textile production thus can not only further the study of Prehispanic technology, but can also help scholars track changing ecological, economic, and social patterns among the agrarian states of Postclassic Central Mexico.
Acknowledgments
Fieldwork by Hirth was supported by the National Science Foundation (INT-7702030), the National Geo­graphic Society (NGS-5092800), and the Wenner-Gren Foundation for Anthropological Research (WGF-3337); Smith was supported by the Wenner-Gren Foundation for Anthropological Research (WGF-3369) and the Graduate College of the University of Illinois. We would like to thank Jorge Angulo and Raul Arana for permission to examine ceramic artifacts from their respective excavations at Teopanzolco and Coatetelco, and the Field Museum of Natural History for permission to study the Frederick Starr collection. Unpublished data and papers were gra­ciously made available by Frances Berdan, Elizabeth Brumfiel, Thomas Charlton, and Mary Parsons. Frances Berdan and Patricia Anawalt kindly provided advice on technical questions concerning Aztec textiles. Some of the data presented here were included in a paper presented by Smith at the 1985 Annual Meeting of the American An­thropological Association in Washington, D.C., and part of the paper was written while Smith was a fellow at Dumbarton Oaks in Washington, D.C. Figures 2-4 were drafted by Patricia Aguirre and Martin Antonio, and Fig­ure 1 by Susan Hirth. Elizabeth Brumfiel provided helpful comments on an earlier draft of this article.
Michael E. Smith is Associate Professor of Anthropology at Loyola University of Chicago. He is currently directing work at rural sites in a provincial area of the Aztec empire, and his methodological interests focus on the archaeological study of complex societies. Mailing address: Department of Sociology
and Anthropology, Loyola University of Chicago, Chicago, IL 60626.
Kenneth G. Hirth is Associate Professor of Anthropology at the University of Kentucky. He has directed archaeological projects in Mexico, Honduras, and Peru, and his research in­terests concern the origin of complex societies and urbanism in Latin America. Mailing address: Department of Anthropol­ogy, University of Kentucky, Lexington, KY 40506.
358 Prehispanic Cotton-spinning Technology/ Smith and Hirth